seismic performance of wuf-w moment connections …earthquake.hanyang.ac.kr/journal/2016/2016,...

Seismic Performance of WUF-W MomentConnections According to Access HoleGeometries

Sang Whan Han,a) M.EERI, Jin Jung,a) and Sung Jin Haa)

The welded unreinforced flange-welded web (WUF-W) moment connectionis an all-welded moment connection, which is one of the prequalified connectionsfor the special moment frame (SMF) specified in AISC 358-10. However, pre-vious experimental studies reported that WUF-W connections with a beam depthof 890 mm failed as a result of brittle beam flange fracture and did not satisfy therequirement for the SMF connection specified in AISC 341-10. In this study, afinite element analysis (FEA) was conducted to investigate the cause of the brittlefailure in the WUF-W connection. This study observed that the brittle failureoccurred in WUF-W connections was mainly attributed to the access hole geo-metry. This study proposed the improved access hole geometry to prevent brittlefailure from WUF-W connections. To verify the analysis results, this study con-ducted an experimental test using one WUF-W connection specimen with animproved access hole geometry. [DOI: 10.1193/060614EQS082M]

INTRODUCTION

In the 1994 Northridge earthquake, many steel moment frames were found to be severelydamaged due to unexpected brittle failures at the welded beam-to-column flange joints(Youssef et al. 1995). To tackle this problem, a number of studies have been conducted(Malley 1998) to propose new moment connections (SAC Joint Venture 2000, Roeder2002, Plumier 1990). In Table 2.1 of AISC 358-10 (2010), seven prequalified moment con-nections for special moment frames (SMF) are specified.

AISC 358-10 includes the welded unreinforced flange-welded web (WUF-W) momentconnection as a prequalified SMF connection, which is an all-welded moment connection.Prequalification of the WUF-W moment connection is based on the results of experimentalresearches from two research groups (Ricles et al. 2002; Lee et al. 2005a, 2005b). Most testedspecimens had excellent connection performance, and satisfied the drift capacity (≥4%) forSMF connections specified by AISC 341-10 (2010). Analytical research (Lu et al. 2000,Mao et al. 2001, El-Tawil et al. 1998) was also conducted for WUF-W connections.This showed that the geometry of the access hole strongly affected the stress distributionin the beam flange of WUF-W connections and their failure pattern.

Han et al. (2014) reported that WUF-W connection specimens with a beam depth of890 mm did not have the satisfactory performance required for SMF connections eventhough the connections were designed and detailed according to AISC 358-10 and

a) Hanyang University, Seoul 133-791, Korea

Earthquake Spectra, Volume 32, No. 2, pages 909–926, May 2016; © 2016, Earthquake Engineering Research Institute909

341-10. Such an unexpected performance has not been reported by previous researchersfor code-complied WUF-W connections with a beam depth not less than 890 mm. Themajor difference between the specimens tested by Han et al. (2014) and Ricles et al.(2002) was the geometry of the weld access hole. Ricles et al. (2002) used the accesshole geometry in their specimens through finite element analysis (FEA) that minimizedthe plastic strain at the toe of the access hole region, whereas Han et al. (2014) used theaccess hole geometry based on configuration parameters within the ranges specified inAISC 341-10.

In this study, FEA was conducted to investigate the cause of the brittle fracture in theWUF-W connection tested by Han et al. (2014). A parametric study was also conducted usingFEA for six WUF-W connections with different access hole geometries. From the results ofFEA, an access hole geometry was proposed, which resulted in the least plastic strain near theaccess hole region. To verify the performance of the WUF-W connection with the proposedweld access hole, this study conducted experimental tests.

DETAILS OF WUF-W MOMENT CONNECTIONS

After the 1994 Northridge earthquake, most new SMF moment connections were devel-oped mainly based on two major strategies: (1) weakening the beam (reduced beam sectionmoment connections), and (2) strengthening the connection (bolted end plate and flange platemoment connections; SAC Joint Venture 2000). Through the application of these two stra-tegies to the moment connection, the plastic hinge moves away from the face of the columnthat is suspected as the location of the brittle fracture.

The WUF-W moment connection was one of the newly developed SMF connections.However, WUF-W connections were not developed based on two strategies mentionedabove. In the WUF-W moment connection, the plastic hinge is not moved away fromthe face of the column. Instead, the WUF-W moment connection adopts design and detailingfeatures intended to permit the connection to achieve the SMF performance criteria withoutbrittle fracture (Lee et al. 2005b). Figure 1a illustrates the WUF-W connection, in which thedetail and welding requirements for the WUF-W connection specified in AISC 358-10 aredenoted. Weld access hole geometry should conform to the requirements of AWS D1.8/D1.8M (2009). A single plate shear connection is provided with a thickness equal to atleast that of the beam web, which is welded to the column flange. The plate is weldedto the beam web with a fillet weld (Figure 1b).

PREVIOUS EXPERIMENTS ON BRITTLE FRACTURED WUF-WCONNECTIONS

Han et al. (2014) conducted experimental tests using four WUF-W connection speci-mens, which satisfied the design and detail requirements specified in AISC 358-10 andAISC 341-10. The columns and beams of all the specimens were made of SS 400(Fy ¼ 235MPa) and SM 490 (Fy ¼ 325MPa) steels, respectively. The specimens also satis-fied proportioning requirements: (1) the clear span-to-depth (l∕d) ratio should exceed a limit-ing value (¼7) for SMF (Section 5.3.1 of AISC 358-10), and (2) the column-beam momentratio (

PM�

pc∕P

M�pb) should be greater than 1 (Section E3.4a of AISC 341-10). The notch-

tough welding metals and welding procedure used for the specimens satisfied the

910 S. W. HAN, J. JUNG, AND S. J. HA

requirements specified in Section J6 of AISC 341-10 and chapter 3 of AISC 358-10. A moredetailed description of the fabrication of the specimens can be found in Han et al. (2014).Figure 1a shows the details of specimen D900-S with a beam depth of 890 mm. Test settingand loading histories are shown in Figures 2 and 3, respectively.

Specimen D900-S experienced an unexpected brittle beam flange fracture near the accesshole during the first positive cycle at a drift ratio of 4%. This specimen did not complete twocycles at a drift ratio of 4% that are required for qualifying SMF connections according toAISC 341-10. However, WUF-W specimens tested by Ricles et al. (2002), which had a simi-lar beam size (¼900mm) and connection details, satisfied the requirements for SMF con-nections. The main difference between the specimens tested by these two studies was theweld access hole geometry, particularly the slope of the access hole; the access hole slopesfor specimens D900-S and the specimens tested by Ricles et al. (2002) were 21° and 13°,respectively. It is noted that the materials (SS400 and SM490 for beams and columns) used inspecimen D900-S satisfied four requirements for the application of structural materials spe-cified in Commentary A.3 of AISC 341-10. The access holes used by Han et al. (2014) weredetermined within the ranges of the configuration parameters specified in AISC-358-10 (seeFigure 1c). To investigate the effect of the access hole geometry on the stress and straindistributions in the WUF-W connection, this study conducted three-dimensional nonlinearfinite element analysis (FEA).

Figure 1. Details of WUF-W moment connection (AISC 358-10 2010).

SEISMIC PERFORMANCEOFWUF-WMOMENT CONNECTIONS ACCORDING TO ACCESS HOLE GEOMETRIES 911

FINITE ELEMENT ANALYSIS

To construct a three dimensional finite element (FE) model and to conduct an analysis,commercial software ABAQUS 6.10.1 (2010), in particular, ABAQUS/CAE and ABAQUS/Standard, was used for modeling and analysis, respectively. The finite element model for theconnection is shown in Figure 2b. In the FE model, the locations of the supports, lateral

Figure 2. (a) Test setup (Han et al. 2014) and (b) finite element (FE) model.

Figure 3. Loading history (Krawinkler et al. 2000).


bracing and loading history were the same as those used in the experimental test (2014).Displacement controlled cyclic loading was applied at the end of the beam, located3.4 m from the column centerline. The loading sequence followed the SAC loading protocol(Krawinkler et al. 2000), as shown in Figure 3.

The results of the FE analysis depend on the size and type of meshes. The three-dimensinal FE model of an eight node brick element can be used. In ABAQUS (2010),three different brick elements are provided: the regular brick element (C3D8), the brick ele-ment with an incompatible deformation mode (C3D8I), and the brick element with reducedintegration (C3D8R). Among those brick elements, element C3D8R is the most efficient interms of computation and storage space because in this element, the reduced integration pro-cedure is used. With an adequate mesh size, FE analysis with C3D8R can produce accurateresults (Lu et al. 2000). In this study, the beam and column flanges were divided into fourmeshes throughout the thickness to obtain accurate analysis results. This study also con-ducted FEA after dividing the beam and column flanges into six meshes throughout the thick-ness, but the change in analysis results was negligible.

The material properties for FE models were obtained from the coupon test results con-ducted by Han et al. (2014). Figure 4 shows the stress strain (σ-ε) curve, and Table 1 sum-marizes the test results. The true stress and strain were obtained by modifying the stress-straincurve (Figure 4) using Equations 1 and 2, which are required for FE modeling in ABAQUSsoftware. The true stress (σtrue) is defined by the stress measured using the true sectional areaof the deformed specimen; true strain (εpl) is the plastic strain, which is obtained by sub-tracting the elastic strain from the measured total strain.

EQ-TARGET;temp:intralink-;e1;62;358σtrue ¼ σð1þ εÞ (1)

EQ-TARGET;temp:intralink-;e2;62;328εpl ¼ lnð1þ εÞ � σ

E(2)

where E is the elastic modulus (205 GPa). A Poisson ratio of 0.3 was assumed for the FEA.

To account for the strain hardening characteristic of the steel material during cyclic tests,this study calibrated the stress-strain relationship obtained from monotonic material tests

Figure 4. Stress-strain curves for steel materials.


based on the results of cyclic material tests conducted by Kaufmann et al. (2001) with similarsteel materials (A36 and A572 Grade 50 steels) as those (SS400 and SM 490) used in Hanet al. (2014). The specified minimum yield strengths (Fy) of A36 and A572 Grade 50 steels,and SS400 and SM 490 steels are 250 MPa and 345 MPa, and 240 MPa and 330 MPa,respectively. In the FE model, the plastic range of the material was idealized by combiningisotropic and kinematic hardenings.

VERIFICATION OF FE MODELING

To verify the accuracy of the FE model used in this study, the test results of specimenD700-S and D900-S were simulated using the FEA, which had beam depths of 680mm and890 mm, respectively. Note that specimen D900-S failed due to brittle beam flange fracturenear the access hole, whereas specimen D700-S did not fail until the end of the experiment. InFigure 5, the hysteretic curves of specimens D700-S and D900-S obtained from FEA were

Table 1. Summary of material test results

Specimen Location material

Yieldstrength(MPa)

Tensilestrength(MPa)

Yieldratio(%)

Elongation(%)

D900-SP21-L69(D900-S)

Beam web SS400 340 482 0.71 22.8

Beam flange SS400 312 468 0.67 23.1Column web SM490 364 535 0.68 21.2Column flange SM490 346 550 0.63 27.0

D900-SP13-L74 Beam web SS400 367 458 0.79 25.2Beam flange SS400 327 437 0.75 31.5Column web SM490 379 541 0.70 24.6Column flange SM490 386 572 0.67 33.2

Figure 5. Hysteretic curves: (a) D700-S, (b) D900-S (D900-SP21-L54).


plotted with those obtained from experiments. The hysteretic curves represent the relation-ship between the forces applied at the beam end and the drift ratios. As shown in Figure 5, thehysteretic curve obtained from the FEA matched the actual hysteretic curve with good accu-racy. Table 2 summarizes the initial stiffness and maximum strength extracted from the hys-teretic curves. The errors in the initial stiffness and maximum strength obtained from the FEAwere 2% and 13%, respectively.

Figure 6 shows the deformed WUF-W connections at a drift ratio of 4%. In this figure,the distribution of von Mises stresses was plotted in the deformed connections. As seen inFigure 6, FEA can predict the shape and location of buckling and the area of yielding withgood accuracy. During the experiment, the area of yielding was detected by visual inspectionof the region where the whitewash fell off.

FEA FOR WUF-W CONNECTIONS WITH DIFFERENT ACCESS HOLEGEOMETRIES

As discussed in previous experimental research, the access hole geometry affects theperformance of WUF-W connections. Mao et al. (2001) reported that the access hole geo-metry strongly affected the stress distribution in the beam flanges and the failure pattern. Inthis study, FEA was conducted to evaluate the inelastic performance of the WUF-W con-nection according to access hole geometries determined within the ranges of the configura-tion parameters permitted in AISC 358-10. The WUF-W connections used in this analyticalstudy were identical to specimen D900-S except for the access hole geometry. This studyconsidered the slope and the overall length of the access hole (Figure 7) as the two majorconfiguration parameters. Two values of the access hole slope were considered, which were13° (Ricles et al. 2002) and 21° (Han et al. 2014). The total lengths of 56 mm, 74 mm, and111 mmwere considered. Thus, this study considered six WUF-W connections with differentaccess hole geometries. Note that all six different geometries of the access hole satisfied therequirements specified in AISC 358-10.

Figure 8 shows the von Mises stress at the beam flange near the column flange of theconnection at a drift ratio of 3%. The von Mises stress was calculated using Equation 3:

Table 2. Test results of D700-S and D900-S

SpecimensInitial stiffness

(KN/mm)Maximum

strength (KN)

D700-S (1) Experiment 14.1 722(2) FEA 14.8 690(1)/(2) 0.95 .05

D900-S(D900-SP21-L69)

(1) Experiment 27.8 1220

(2) FEA 27.3 1058(1)/(2) 1.02 1.13


EQ-TARGET;temp:intralink-;e3;41;306σeff ¼ffiffiffiffiffiffiffiffiffiffiffiffiffi2

3SpijS

pij

r(3)

where Spij is the deviatoric stress components (Spij ¼ σij þ σmδij), σij is the Cauchy stress com-ponents, and i and j represent the global coordinates (i ¼ 1, 2, 3 and j ¼ 1, 2, 3), σm is thehydrostatic stress, and δij is the value of Kronecker delta.

As shown in Figure 8, the maximum stress was detected approximately at 90 mm fromthe column face irrespective of the total length of the access hole and the location of theaccess hole toe, which is the location of the local buckling in the beam flange. SpecimensD900-SP13-L56 and D900-SP13-L74 had a total access hole length of 56 and 74 mm,respectively, and both specimens had an access hole slope of 13°, where “SP” and “L”stand for the access hole slope and total access hole length, respectively (Figure 7). Asshown in Figure 7a and 7b, the locations of the access hole toe of D900-SP13-L56 andD900-SP13-L74 were located at 0 mm and 18 mm from the column face, respectively.Those locations were not close to the location of the maximum stress (¼90 mm fromthe column face), which are shown in Figure 8a and 8b. However, in specimen D900-SP13-L111, the location of the access hole toe (¼54 mm from the column face) wasclose to the location of maximum stress (Figure 8c). This resulted in a sharp increase in

Figure 6. Specimens (a) D700-S and (b) D900-S at the first loading cycle at 4% drift ratio.


the level of the maximum stress. The level of the maximum stress in D900-SP13-L111 wasmuch higher than that of D900-SP13-L56 and D900-SP13-L74. In Figure 8, the darker areaindicates the portion with greater stress. Such stress may increase the potential of a beamflange fracture near the access hole.

Figure 7. Access hole geometries considered in this study.

Figure 8. von Mises stress distribution at the beam bottom flange at 3% drift ratio.


To investigate the effect on the slope of the access hole on stress distribution near accesshole, Figure 9 was plotted, which shows the stress distributions in specimens D900-SP13-L74 and D900-SP21-L74. Even though the slopes of the access holes in these specimenswere different (13° and 21°), the total access hole length was the same (¼74 mm; Figure 7band 7d). Figure 9 shows the distribution of the von Mises stress in the beam flange of thesetwo specimens at a drift ratio of 3%. In specimen D900-SP13-L74, the maximum x- andy-directional stresses were detected in the beam flange near the access hole toe and inthe beam web along the perimeter of the access hole, respectively. The peak vonMises stress,which is the combined stress of the x- and y-directional stresses, was detected in the beamweb along the access hole perimeter (Figure 9). However, in specimen D900- SP21-L74,which had an access hole slope of 21°, the maximum values of the x- and y-directional stres-ses and von Mises stresses were detected at the same location in the beam flange at the accesshole toe. This indicates that with an increase in the access hole slope, the location of themaximum stresses moves from the access hole perimeter to the access hole toe, resultingin the greater potential of a brittle beam flange fracture at the access hole toe.

This study also attempted to find the access hole geometry that has the least plastic straindemand in the beam flange near the access hole toe. Six models having six different accesshole geometries were considered as shown in Figure 7. At a drift ratio of 3%, the PEEQ indexwas estimated using Equation 4 for each specimen. The PEEQ index was also used in Maoet al. (2001).

EQ-TARGET;temp:intralink-;e4;41;389PEEQ Index ¼ PEEQ∕εy (4)

Figure 9. von Mises stress distribution at the access hole region at 3% drift ratio.


where PEEQ is the effective plastic strain, and εy is the yield strain. The PEEQ can be cal-culated using Equation 5:

EQ-TARGET;temp:intralink-;e5;62;394PEEQ ¼ffiffiffiffiffiffiffiffiffiffiffiffi2

3εpijε

pij

r(5)

where εpij is the plastic strain of global coordinates, ði; jÞ.Specimen D900-SP21-L111 had the largest PEEQ index among the six models. Figure 10

shows the maximum value of the PEEQ index of each specimen, normalized by that of speci-men D900-SP21-L111. Specimen D900-SP13-L74, which had an access hole slope of 13°and a total length of 74 mm, had the smallest PEEQ index among the specimens.

EXPERIMENTAL TEST FOR WUF-W CONNECTION WITH AN IMPROVEDACCESS HOLE

In this study, experimental tests were conducted for two reasons: (1) to confirm that thebrittle beam flange fracture observed in specimen D900-S [15] was mainly due to the accesshole geometry, and (2) to verify whether specimen D900-SP13-L74 with the improved accesshole geometry had the satisfactory performance required for SMF connections. Note that theaccess hole geometries of specimens D900-S (renamed D900-SP21-L69 hereafter) andD900-S-SP13-L74 satisfied the requirements for the SMF connection specified in AISC358-10 and AISC 341-10. Both specimens were identical except for the access hole geome-try. Tests were conducted using the same test setting (Figure 2) and loading history (Figure 3)used for D900-SP21-L69. The access hole geometries of both specimens are summarized inTable 3.

The stress-strain curves for specimen D900-S-SP13-L74 obtained from coupon tests areplotted in Figure 4. From the stress strain curves, the mechanical properties of steel materialused for the specimen is summarized in Table 1. From the test results, it was shown that the

Figure 10. Maximum normalized PEEQ index at the critical location at 3% drift ratio.


steel materials, SS 400 and SM 490, used in this experiment satisfied the requirement for thesteel material of SMF members specified in ANSI/AISC 341-10.

To detect the yielding process in the critical regions of the connection, the specimenwas white washed before the experiment. Linear variable differential transformersand strain gauges were placed in the connection specimen according to Ricleset al. (2002).

HYSTERETIC BEHAVIORS

Figure 11 shows the hysteretic curves of D900-SP13-L74, representing the relationshipbetween the drift ratio (θ) and beam end moment at the column face normalized by the beamplastic moment (Mp). Table 4 summarizes the important values extracted from the hystereticcurves.

Figure 12 shows the deformed WUF-W connection specimens at different loadingstages. Specimen D900-SP21-L69 first experienced yielding in the beam flanges nearthe column flange at drift ratios of 0.5% (Figure 12a). At a drift ratio of 3%, beam flangelocal buckling was detected in D900-SP21-L69 (Figure 12b), but noticeable strengthdeterioration was not observed in this specimen due to the local buckling (Figure 12b).During the first loading cycle at a drift ratio of 4%, significant local buckling occurred inthe bottom flange of the beam, and the strength sharply dropped (Figure 12c); this wasfollowed by a sudden fracture of the beam bottom flange at the access hole toe(Figure 12d).

In specimen D900-SP13-L74, the first yielding was detected in the top beam flange andthe beam near the column flange at drift ratios of 0.75% (Figure 12e) and 1.5%, respectively.At a drift ratio of 3%, beam flange local buckling was first detected in the beam flange of thebeam (Figure 12f). After a drift ratio of 3%, the connection strength deteriorated due to

Table 3. Weld access hole geometry

Specimen a (mm) b (mm) c (mm) d (mm) e (mm) f (mm) g (°)

D900-SP21-L69 (D900-S) 23.0 69.0 23.0 23.0 10.0 33.0 21°D900- SP13-L74 18.0 74.0 23.0 20.0 10.0 30.0 13°


Figure 11. Hysteresis curves: (a) D900-SP21-L69 (D900-S), (b) D900-SP13-L74.

Table 4. Summary of test result

SpecimenD900-SP21-L69

(D900-S) D900-SP13-L74

Total story drift (rad) 0.030 0.051Total plastic rotation (rad) 0.019 0.045Panel zone maximum plastic rotation (rad) 0.004 0.002Beam maximum plastic rotation (rad) 0.016 0.048Column maximum plastic rotation (rad) 0.001 0.001Mf ∕Mp 1.44 1.34Location of fracture Beam flange at

access hole toeNot observed


significant local buckling in the beam flange. At drift ratios of 4% and 5%, the connectionstrength decreased to 87% and 64% of the beam plastic moment (Mp), respectively(Figure 12g and 12h). The experiment was terminated at a drift ratio of 5% because ofthe safety of the lateral supports placed for the specimen.

EFFECT OF ACCESS HOLE GEOMETRY

Specimen D900-SP21-L69 (Han et al. 2014) had total drift and plastic drift ratios of 3%and 1.9%, respectively, which did not satisfy the requirement for the SMF connection spe-cified in AISC-358-10. On the other hand, specimen D900-SP13-L74, which had animproved access hole geometry, produced total and plastic drift ratios of 5% and 4.5%,respectively; at a drift ratio of 4%, the connection strength was reduced to 87% of its max-imum strength (Mmax), which satisfied the SMF connection requirements specified inAISC341-10. Considering that both specimens were identical in dimension, member sec-tions, material properties, and the fabrication process except for access hole geometry,the difference in connection performance between the two specimens was mainly inducedby the access hole geometry. This observation confirmed the results of FEA in the previoussections.

To investigate the effect of access hole geometry on the strain demands in the beamflanges, this study compared strains in the beam top flanges of specimens D900-SP21-L69 and D900-SP13-L74 at a drift ratio of 4%. The strains were measured by strain gaugesplaced at 50 mm from the column face and were plotted in Figure 13. Based on the experi-mental results, the strain in specimen D900-SP13-L74 was 18–43% smaller than that in spe-cimen D900-SP21-L69. As shown in Figure 13, even though strain results obtained fromFEA were somewhat larger than those obtained from experiments, strains in specimen

Figure 12. Fracture of specimens: (a–d) D900-SP21-L69 and (e–h) D900-SP13-L74.


D900-SP13-L74 was 15–40% smaller than those in specimen D900-SP21-L69. This indi-cates that a specimen having the improved access hole geometry had less potential for brittlebeam flange fracture.

ENVELOP CURVES AND CUMULATIVE DISSIPATED ENERGY

The envelope curves are plotted in Figure 14a for specimens D900-SP21-L69 and D900-SP13-L74, which represent the relationship between the drift ratio and beam moment at thecolumn face. These specimens had identical elastic curves. After a drift ratio of 1%, thestrength of both specimens deteriorated. The post yield stiffness of specimen D900-SP21-L69 was larger than that of specimen D900-SP13-L74.

The strength of specimens D900-SP21-L69 and D900-SP13-L74 increased up to a driftratio of 3% and 2%, respectively, and their maximum strengths were 3,930 kN·m (¼1.44)and 3,630 kN·m (¼1.34), respectively.

Specimen D900-SP21-L69 failed due to the beam flange fracture without completing onecycle at a drift ratio of 4% (Figure 11a), while specimen D900-SP13-L74 did not experiencefailure until reaching a drift ratio of 5%.

The dissipated cumulative energy of specimens D900-SP21-L69 and D900-SP13-L74was calculated, and was plotted with respect to drift ratios in Figure 14b. Until a driftratio of 3%, both specimens dissipated almost the same amount of energy. The total cumu-lative energy of specimens D900-SP13-L74 (¼1;900 kJ) is twice that of specimen D900-SP21-L69 (¼900 kJ). This was quite an obvious result because specimen D900-SP21-L69 failed in a much earlier loading stage compared with D900-SP13-L74. Consideringthat specimen D900-SP13-L74 is identical to D900-SP21-L69 except for the access holegeometry, the energy dissipation capacity of WUF-W connections can be significantlyenhanced by using an improved access hole geometry.

Figure 13. Strain demands on the beam flange at the first loading cycle at 4% drift ratio.


CONCLUSION

This study conducted FEA to investigate the cause of brittle failure of pre-qualifiedWUF-W connections designed and detailed according to AISC 358-10 and AISC 341-10in order to propose an improved access hole geometry. To verify the FEA results, experi-mental tests were conducted using the connection specimen with improved access hole geo-metry. Based on the analytical and experimental results, the following conclusions andrecommendations were made:

1. This study showed that by using FEA, the hysteretic curves of WUF-W specimens(specimens D900-SP21-L69 and D700-SP13-L74) were accurately simulated.

Figure 14. (a) Envelop curves and (b) cumulative energy dissipation for connection specimens.


Based on the FEA results of the WUF-W connections, the cause of the fracturedWUF-W specimen (D900-SP21-L69) was attributed to the access hole geometry.

2. From FEA, it was observed that the location of local buckling in the beam flange ofWUF-W connections did not vary according to the different geometries of the accessholes. When the access hole toe coincided with the location of beam flange localbuckling, the maximum stress in the beam flange was sharply increased, which wassuspected to have caused the brittle beam flange fracture in the early loading stage.

3. After conducting the FEA for WUF-W connection specimens with six differentaccess hole geometries, the PEEQ indices were compared. The specimen withan access hole slope (SP) of 13° and a total length (L) of 74 mm had the smallestPEEQ index among the six specimens.

4. To verify the improvement in the hysteretic behavior according to the change in theaccess hole geometry of WUF-W connection specimens, an experimental test wasconducted using specimen D900-SP13-L74, which was identical to specimen D900-SP21-L69 except for the access hole geometry. Specimen D900-SP13-L74 had totaland plastic drift capacities of 5% and 4.4%, respectively, which satisfied the require-ment for the SMF connection specified in AISC 358-10. By using the improvedaccess hole geometry, the strain in the beam flange near the access hole was reducedas much as 43%.

5. Specimen D900-SP13-L74, with the improved access hole geometry, had an energydissipation capacity twice that of specimen D900-SP21-L69.

6. Based on the analytical and experimental studies, some WUF-W connections withan access hole geometry determined using the configuration parameters within theranges permitted by AISC 358-10 may not guarantee the satisfactory performancerequired for SMF connections. The proper ranges of the configuration parametersneed to be proposed for the weld access holes of WUF-W connections in futureresearch.

ACKNOWLEDGMENTS

Authors acknowledge the financial supports provided by the National ResearchFoundation of Korea (No. 2014R1A2A1A11049488). Soon Wook Cho, a graduate studentat Hanyang University, Seoul, Korea, assisted in the massive computing effort. Finally, thevaluable comments offered by two anonymous reviewers are greatly appreciated.

REFERENCES

American Institute of Steel Construction (AISC), 2010. Seismic Provisions for Structural SteelBuildings, ANSI/AISC 341-10, Chicago, IL.

American Institute of Steel Construction (AISC), 2010. Prequalified Connections for Special andIntermediate Steel Moment Frames for Seismic Applications, ANSI/AISC 358-05, Chicago, IL.

American Welding Society (AWS), 2009. Structural Welding Code-Seismic Supplement, AWSD1.8/D1.8M, Miami, FL.

Dassault Systemes, 2010. ABAQUS Analysis User’s Manual, Providence, RI.El-Tawil, S., Mikesell, T., Vidarsson, E., and Kunnath, S. K., 1998. Strength and Ductility

of FR Welded-Bolted Connections, Rep. No. SAC/BD-98/01, SAC Joint Venture, Richmond,CA.


Han, S. W., Moon, K. H., and Jung, J., 2014. Cyclic performance of welded unreinforced flange–welded web moment connections, Earthquake Spectra 30, 1663–1681.

Kaufmann, E. J., 2001. Characterization of Cyclic Inelastic Strain Behavior on Properties ofA572 Gr.50 and A913 Gr. 50 Rolled Sections, ATLSS Rep. No. 01-13, Bethlehem, PA.

Krawinkler, H., Gupta, A., Medina, R., and Luco, N., 2000. Loading Histories for Seismic Per-formance Testing of SMRF Components and Assemblies, Rep. No. SAC/BD-00/10, SAC JointVenture, Sacramento, CA.

Lee, D., Cotton, S. C., Hajjar, J. F., Dexter, R. J., and Ye, Y., 2005a. Cyclic behavior of steelmoment-resisting connections reinforced by alternative column stiffener details I: Connectionperformance and continuity plate detailing AISC Engineering Journal 42, 189–214.

Lee, D., Cotton, S. C., Hajjar, J. F., Dexter, R. J., and Ye, Y., 2005b, Cyclic behavior of steelmoment-resisting connections reinforced by alternative column stiffener details II: Panel zonebehavior and doubler plate detailing, AISC Engineering Journal 42, 215–238.

Lu, L. W., Ricles, J. M., and Fisher, J. W., 2000. Critical issues in achieving ductile behavior ofwelded moment connections, Journal of Constructional Steel Research 55, 325–341.

Malley, J., 1998. SAC steel project: Summary of phase 1 testing investigation results. Engineer-ing Structures 20, 300–309.

Mao, C., Ricles, J., Lu, L. W., and Fisher, J., 2001. Effect of local details on ductility of weldedmoment connections, ASCE Journal of Structural Engineering 127, 1036–1044.

Plumier, A., 1990. New idea for safe structure in seismic zones, Proceedings of IABSE Sympo-sium, Belgium, Brussels, 253–258.

Ricles, J. M., Mao, C., Lu, L. W., and Fisher, J. W., 2002. Inelastic cyclic testing of weldedunreinforced moment connections, ASCE Journal of Structural Engineering 128, 429–440.

Roeder, C. W., 2002. Connection performance for seismic design of steel moment frames, ASCEJournal of Structural Engineering 128, 300–309.

SAC Joint Venture, 2000. Recommended Seismic Design Criteria for New Steel Moment FrameBuildings, FEMA 350, Richmond, CA.

Youssef, N. F. G., Bonowitz, D, and Gross, J. L., 1995. A survey of steel moment resisting framebuildings affected by the 1994 Northridge earthquake, Rep. No. NISTIR-5625, National Insti-tute of Standard and Technology, Gaithersburg, MD.

(Received 6 June 2014; accepted 13 July 2015)


http://dx.doi.org/10.1016/S0143-974X(99)00092-9

http://dx.doi.org/10.1016/S0141-0296(97)00033-3

http://dx.doi.org/10.1016/S0141-0296(97)00033-3

http://dx.doi.org/10.1061/(ASCE)0733-9445(2001)127:9(1036)




seismic performance of wuf-w moment connections …earthquake.hanyang.ac.kr/journal/2016/2016,...

Documents