Reliability Analysis of Concentrically Loaded Fillet Welded Joints

Transcription

1 University of Alberta Department of Civil & Environmental Engineering Strctral Engineering Report No. 27 Reliability Analysis of Concentrically Loaded Fillet Welded Joints by Chnlong Li Gilbert Y. Grondin and Robert G. Driver October, 2007

2 University of Alberta RELIABILITY ANALYSIS OF CONCENTRICALLY LOADED FILLET WELDS Strctral Engineering Report 27 by Chnlong Li Gilbert Y. Grondin Robert G. Driver DEARTMENT OF CIVIL AND ENVIRONMENTAL ENGINEERING Edmonton, Alberta October, 2000

3 ABSTRACT Tests on joints of crciform configration condcted in a previos test program indicated that fillet welds in these joints may possess redced strength and dctility compared to transverse welds in lapped joints. An experimental program consisting of 2 fillet weld crciform specimens was condcted to investigate effect of root notch orientation on weld strength and dctility. The reliability analyses presented in the previos phases of this program were condcted sing mostly test reslts obtained at the University of Alberta. To agment this database of test reslts, test data from varios test programs on lapped splices with single and mltiple orientation welds and on crciform test specimens were collected. This provides a larger database of test reslts, which better reflect the variation present in indstry. A reliability analysis was condcted to assess the crrent North American and proposed design eqations for lapped joints with single and mltiple orientation welds and for crciform connections. ii

4 ACKNOWLEDGEMENTS Fnding for this research project was provided by the Natral Sciences and Engineering Research Concil of Canada and the American Institte of Steel Constrction. Fnding was also provided by Canadian Institte of Steel Constrction in the form of the G. L. Klak scholarship to the senior athor. The tests presented in this report were condcted by Mr. Logan Callele, Mr. Andre Blanchard and Mr. Derek Kramar, from the Department of Civil and Environmental Engineering of the University of Alberta. iii

5 TABLE OF CONTENTS. INTRODUCTION.... Backgrond....2 Objectives and Scope Units Used in this Report LITERATURE REVIEW Introdction Fillet Weld Strength as a Fnction of Loading Direction Strength of Crciform Fillet Weld Specimens Load of Joints with Mltiple Orientation Fillet Welds (MOFW) Effects of Weld Sizes and Lengths on Fillet Weld Strength Smmary EXERIMENTAL ROGRAM AND RESULTS Introdction Test arameters Specimens Description re-test Measrements Instrmentation and Test rocedre Main lates Misalignment Second Order Effects Crciform Specimen Capacities Weld Strains Calclated from Deformations Measred with LVDTs Effect of Root Notch Orientation on Weld Strength Effect of Root Notch Orientation on Dctility ANALYSIS OF TEST RESULTS Introdction Smmary of Test Data from Different Sorces...50 iv

6 4.2. Geometric Factor, G Material Factor, M Material Factor, M rofessional Factor,, for Joints with Only One Weld Orientation rofessional Factor,, for Joints with Mltiple Weld Orientations rofessional Factor,, for Crciform Connections Safety Level of Lapped Splice Connections Connections with Single Orientation Fillet Welds Connections with Fillet Welds Oriented in Mltiple Directions Safety Level of Crciform Connections SUMMARY, CONCLUSIONS AND RECOMMENDATIONS Smmary Conclsions Recommendations for Ftre Research...85 REFERENCES...87 AENDIX A ALL-WELD-METAL TENSION COUON TESTS...9 AENDIX B FILLET WELD SECIMEN MEASUREMENTS...96 AENDIX C SECIMEN RESONSE CURVES (ROOM TEMERATURE TESTS)...04 AENDIX D RESULTS OF WELD TESTS AT LOW TEMERATURE...20 AENDIX E RESULTS OF OTHER TEST ROGRAMS...29 AENDIX F NEW SECIMEN DESIGN DRAWINGS...25 AENDIX G RESULT OF ADDITIONAL SIX SECIMENS TESTED IN HASE v

7 LIST OF TABLES Table 3. Test Matrix Crciform Specimens...24 Table 3.2 Mechanical roperties of Weld Metal...24 Table 3.3 Smmary of re-test Weld Measrements...25 Table 3.4 Measred Eccentricities in the Crciform Specimens...26 Table 3.5 Bending Effect on Crciform Specimens...27 Table 3.6 Smmary of Test Capacities Obtained at Room Temperatre...28 Table 3.7 Ultimate and Fractre Weld Deformations and Strains...28 Table 3.8a Comparison of Fillet Weld Strength (2.7 mm, E70T 7)...29 Table 3.8b Comparison of Fillet Weld Strength (2.7 mm, E7T8 K6)...30 Table 3.8c Comparison of Fillet Weld Strength (6.4 mm, E70T 4)...3 Table 3.8d Comparison of Fillet Weld Strength (6.4 mm, E70T7 K2)...32 Table 3.9 Smmary of Strength...33 Table 3.0a Comparison of Fillet Weld Strains (2.7 mm, E70T 7)...34 Table 3.0b Comparison of Fillet Weld Strains (2.7 mm, E7T8 K6)...35 Table 3.0c Comparison of Fillet Weld Strains (6.4 mm, E70T 4)...36 Table 3.0d Comparison of Fillet Weld Strains (6.4 mm, E70T7 K2)...37 Table 4. Table 4.2 Table 4.3 Smmary of Geometric Factor G from Varios Sorces...57 Geometric Factor G for Specimens from Crrent Test rogram...59 Smmary of Geometric Factor G from hase throgh Table 4.4 Smmary of Material Factor M...6 Table 4.5 Smmary of Material Factor M 2 per Eqation 4.6a...62 Table 4.6 Smmary of rofessional Factor for SOFW Joints...63 Table 4.7a Smmary of rofessional Factor for MOFW Joints for Eqation Table 4.7b Smmary of rofessional Factor for MOFW Joints for Eqation vi

8 Table 4.7c Smmary of rofessional Factor for MOFW Joints for Eqation 4.a and b...67 Table 4.8 rofessional Factor for Crciform Specimens from hase and Table 4.9 Smmary of rofessional Factor for Crciform Connections...69 Table 4.0 Smmary of Safety Indices for SOFW Joints...70 Table 4. Smmary of Safety Indices for MOFW Joints...7 Table 4.2 Smmary of Safety Indices for SOFW Specimens with Ot-of-plane Eccentricity from Ligtenberg (968)...72 Table 4.3 Smmary of Safety Indices for Crciform Joints...73 vii

9 LIST OF FIGURES Figre 3. Dimensions of Crciform Test Specimens...38 Figre 3.2 Definition of Fillet Weld re-test Measrements...39 Figre 3.3 Eccentricity Measrements...40 Figre 3.4 Weld Deformation Measrements...4 Figre 3.5 Arrangement of Strain Gages...42 Figre 3.6 Load Eccentricity and Bending Effect...43 Figre 3.7 Effect of Root Notch Orientation on Weld Strength...43 Figre 4. Variation of G andv G with Weld Size...74 Figre 4.2 Vale of M as a Fnction of the Nominal Tensile Strength from Varios Sorces...75 Figre 4.3 Variation of M in Varios Data Sets...75 Figre 4.4 Werner Specimen...76 Figre 4.5 Bias Coefficient M 2 as a Fnction of Weld Size...76 Figre 4.6 Effect of Tensile Strength on the Bias Coefficient M 2 for All Test Specimens in Table Figre 4.7 rofessional Factor for Transverse Weld verss Measred Weld Size...77 Figre 4.8 rofessional Factor for Transverse Weld verss Measred Filler Figre 4.9 Metal Tensile Strength...78 MOFW Specimen with Ot-of-plane Eccentricity from Ligtenberg (968)...79 Figre 4.0 rofessional Factor for Combined Transverse and Longitdinal Welds verss Measred Filler Metal Tensile Strength...80 viii

10 LIST OF SYMBOLS A throat Measred throat area of the weld (mm 2 ) A w Theoretical throat area of the weld (mm 2 ) C CL Adjstment factor for the weld resistance factor The central plate leg is the fillet weld leg that is on the central plate of the crciform specimen CRF( θ ) Combination redction factor for any fillet weld orientation, θ. d LL MOFW ML Average measred fillet weld leg dimension The lap plate leg is the fillet weld leg that is on the lap plate of the lapped specimen Mlti-orientation fillet weld The main plate leg is the fillet weld leg that is on the main plate of the specimen MTD Measred throat dimension, see definition in Section 4. Normalized The vale of the term ( A throat A of the filler metal ST throat ) divided by the ltimate tensile strength Applied load ST Maximm applied static load Maximm applied load of a weld with a loading angle of θ s, s 2 Fillet weld leg dimensions SOFW V G Single orientation fillet weld Coefficient of variation for the measred-to-nominal weld dimension ix

11 V M Coefficient of variation for the measred-to-nominal ltimate tensile strength of the filler metal V M 2 Coefficient of variation for the measred-to-predicted ltimate shear strength of the filler metal V Coefficient of variation for the test-to-predicted weld capacity V R Coefficient of variation for the resistance of the weld V r of a fillet weld connection X Nominal ltimate tensile strength of the filler metal (Ma) α R β Δ Δ f Coefficient of separation Reliability index Fillet weld deformation Fillet weld deformation at weld fractre Δ lt Fillet weld deformation at ltimate load ε ε y Strain measred by strain gage Strain eqal to θ λ μ G Angle between the axis of the fillet weld and the direction of the applied load (degrees) Regression coefficient sed to characterize fillet weld response Regression coefficient sed to characterize fillet weld response Bias coefficient for the theoretical weld dimension M Bias coefficient for the ltimate tensile strength of the filler metal M 2 Bias coefficient for the ltimate shear strength of the filler metal x

12 Mean test-to-predicted weld capacity R Bias coefficient for the resistance of the weld σ The tensile strength of the weld metal τ Measred ltimate shear strength for a longitdinal weld (Ma) φ φ w Resistance factor Weld resistance factor Symbols sed in Appendix E may be different from this list and were explained where they appeared in Appendix E. xi

13 CHATER INTRODUCTION. Backgrond The effect of load orientation on the strength and dctility of fillet welds has been recognized for a long time (Btler and Klak, 97; Clark, 97 and Miazga and Kennedy, 989). Transverse fillet welds, loaded at a right angle to the axis of the weld, provide the most strength bt least dctility, whereas longitdinal welds provide the least strength, bt most dctility. Relationships between load orientation and strength and dctility have been adopted in North American design standards (CSA, 200; AISC, 2005). The crrent design eqation for strength of fillet welds sed in the North American design standards originated from the work of Miazga and Kennedy (986, 989) and Lesik and Kennedy (990). The test specimens by Miazga and Kennedy were prepared sing the shielded metal arc welding (SMAW) process and an E484 filler metal, which has no toghness reqirement. This work was recently expanded by Ng et al. (2002), Deng et al. (2003), and Callele et al. (2005) to inclde the more prevalent flx-cored arc welding (FCAW) process and filler metals both with and withot a toghness reqirement. The reslts demonstrated that the crrent design eqation, which acknowledges a strength for transverse welds 50% higher than that for longitdinal welds, provides an adeqate level of safety for connections with fillet welds loaded in the plane of the joint. In the case of a connection that combines welds with different orientations, the dctility of each weld segment mst be acconted for when the strength of the joint is evalated. In a typical welded joint that combines weld segments of different orientations, the weld segments with the least dctility sally develop their fll strength, whereas the segments with the most dctility develop only part of their strength. Klak and Grondin (2003) observed that the only 85% of the strength of longitdinal welds is developed when combined with transverse welds in the same joint. This observation was made from tests

14 on joints that combined transverse welds, longitdinal welds, and bolts in the same shear plane. Callele et al. (2005) frther expanded the research work of Ng et al. (2002) and Deng et al. (2003) to inclde connections with mltiple orientation fillet welds (MOFW). It was demonstrated that the strength of differently oriented weld segments is not flly mobilized at failre of the welded joint. They confirmed the observation of Klak and Grondin (2003) that when longitdinal welds are combined with a transverse weld in the same joint, only 85% of the strength of the longitdinal weld is mobilized at failre of the joint. Based on their test reslts, Callele et al. (2005) proposed a reliability based procedre to accont for the effect of any weld orientation on dctility and contribtion to the strength of MOFW. The weld research program at the University of Alberta has inclded six different electrode classifications, which represents only a small portion of the electrode classifications sed in indstry. It is therefore desirable to condct an extensive review of the literatre to agment the database of test reslts on welded concentrically loaded joints..2 Objectives and Scope This report presents reslts of the forth phase of a research project initiated at the University of Alberta to investigate the behavior of concentrically loaded fillet welded connections. Ng et al. (2002), Deng et al. (2003) and Callele et al. (2005) presented the reslts of previos three phases of the project. The forth phase presents additional test reslts from welded crciform joints. Tests on joints of this configration condcted in phase indicated that fillet welds in these joints may possess redced strength and dctility compared to transverse welds in lapped joints. In order to meet the objective of this project, an experimental program consisting of 2 fillet weld crciform specimens was condcted. All specimens were prepared sing the FCAW process and two electrodes, namely, E70T-7 and E7T8-K6. The E70T-7 filler metal has no toghness reqirement, while the E7T8-K6 filler metal has a toghness reqirement of 20 J at -29ºC (AWS 998). 2

15 The reliability analyses presented in the previos phases of this program were condcted sing test reslts obtained primarily at the University of Alberta. Another main objective of this phase is to agment this database of test reslts to inclde test reslts from other sorces. This provides a larger database of test reslts that better reflects the variation present in indstry. Test data from varios test programs on lapped splices with single and mltiple orientation welds and on crciform test specimens were collected. A reliability analysis was condcted to assess the crrent North American design eqations..3 Units Used in this aper SI nits are generally sed throghot this docment, althogh there are exceptions. The filler metal designations se imperial nits as implemented in the AWS classification. This exception was made becase the AWS classification is more commonly sed in indstry. The weld sizes are the direct conversion from their imperial nits becase of the practice adopted by the fabrication workshop. 3

16 CHATER 2 LITERATURE REVIEW 2. Introdction Althogh both analytical and experimental investigations of the strength and dctility of fillet welds exist, this review covers only experimental research. The literatre review focses particlarly on experimental work condcted from the 960s to the present on concentrically loaded fillet weld connections, as this is the most relevant to the crrent research. A detailed literatre review of fillet weld research is available from Ng et al. (2002). The test data of some experimental programs are reprodced and analyzed in Appendix E, and the reslts of those tests are sed in Chapter 4 for a reliability analysis. 2.2 Fillet Weld Strength as a Fnction of Loading Direction Ligtenberg (968) presented the reslts of an international test series on doble lapped splice fillet weld joints loaded in tension. Inclding Canada, contries participated in this stdy and each contry performed separate tests nder the direction of a common committee. The specimen configrations inclded longitdinal welds only, transverse welds only, and combined longitdinal and transverse welds with different proportions of weld throat size and weld length. The majority of the test series were performed on St.37 steel (minimm ltimate tensile strength 360 Ma), or material of similar qality, and a few additional test series were carried ot on St.52 steel (minimm ltimate tensile strength 50 Ma) to broaden the scope of application of the test reslts. To diversify the electrodes and welding processes, each contry selected three electrodes (acid coated, basic and rtile) in sch a way that they were made and commonly sed in the participating contry. The weld metal tensile strength ranged from abot 450 Ma to 580 Ma and the weld throat size ranged from 3 mm to 0 mm. The strength of transverse welds was fond to be.6 times that of longitdinal fillet welds when the weld 4

17 strength was defined as one half of the sm of steel plate ltimate tensile strength and the filler metal ltimate tensile strength. Based on the test reslts presented by Ligtenberg (968), Bornscheer and Feder (966) designed a test program to investigate frther the effect on fillet weld strength of weld length, weld throat dimension, and the ratio of the area of the plate to the cross-sectional area of the fillet weld. The specimen configrations were the same as those reported by Ligtenberg (968) and a rtile-type electrode was sed. The throat dimensions of specimens were 4 mm, 8 mm and 2 mm. The ratio of the strength of transverse welds to that of longitdinal welds was fond to be similar to the reslts presented by Ligtenberg (968). After the international test series presented by Ligtenberg (968), Kato and Morita (969) reported tests that were designed to frther investigate the effects of weld metal mechanical properties, the amont of weld penetration, and weld leg sizes on fillet weld strength. Basic and rtile electrodes were sed. The weld leg size ranged from 5 mm to 40 mm for transverse welds and from 5 mm to 22 mm for longitdinal welds. Kato and Morita proposed that the strength of transverse fillet welds was statistically.46 times that of longitdinal fillet welds. The test reslts reported by Ligtenberg (968) were also sed to sbstantiate their strength prediction formlae. Higgins and reece (969) reported a series of 68 tests on doble lapped splice specimens with all longitdinal or all transverse fillet welds. A variety of shielded metal arc welding electrodes (E60XX, E70XX, E90XX and E0XX), base metals (ASTM A36, A44 and A54), fillet weld sizes (6.35 mm, 9.53 mm and 2.7 mm), and weld lengths (from 38. mm to 0.6 mm) were sed. Another important factor considered in the design of the test specimens was the matching of base metal and the weld metal. For fillet welds made with E70XX electrodes and deposited on matching base metals of grade A44 steel, the strength of transverse welds was fond to be.57 times that of longitdinal welds. For fillet welds made with E0X electrodes and deposited on matching base metals of grade A54 steel, the average strength of transverse welds was.44 times that of longitdinal welds. 5

18 Clark (97) reported a series of 8 tests condcted to investigate the variation of strength and dctility with the loading direction. The specimens inclded 0º (longitdinal), 30º, 60º and 90º (transverse) fillet weld orientations. The reslts showed an increase in strength of approximately 70% as the loading angle changed from 0º to 90º. Btler and Klak (969, 97) reported a series of 23 concentrically loaded doble lapped connections with 6.35 mm (¼ in.) fillet welds of different angles, namely 0º (longitdinal), 30º, 60º and 90º (transverse). The objective of the tests was to establish load deformation relations for elemental lengths of fillet weld. The reslts showed that both weld strength and deformation capacity vary with the loading direction. The specimens were prepared sing E60XX electrodes and CSA G40.2 steel plates, which have a specified yield strength of 300 Ma and a minimm tensile strength of 450 Ma. The test reslts showed that the increase in strength of the fillet welds was approximately 44% as the angle of loading changed from 0º to 90º. Swannell and Skews (979a, b) reported the reslts of a series of tests designed to obtain load deformation relationships for different loading angles. The main specimens were loaded in compression, which differs from most of other test programs. The major tests consisted of six specimens loaded in compression for every loading angle of 0º (longitdinal), 30º, 60º and 90º (transverse). The specimens were prepared sing E603 electrodes and steel plates with a minimm yield strength of 20 Ma and a minimm tensile strength of 40 Ma. The leg size was 6.35 mm. The test reslts showed an increase in strength of approximately 9% as the angle of loading changed from 0º to 90º. Miazga and Kennedy (986, 989) reported the reslts of tests on 42 fillet weld doble lapped splice specimens loaded in tension. The loading angle varied from 0º to 90º in 5º intervals. Two weld sizes were tested, namely, 5 mm and 9 mm, deposited sing the SMAW process with E704 electrodes. The ratio of the transverse weld strength to the longitdinal weld strength was.28 for the 5 mm welds and.6 for the 9 mm welds, with an average of.43 for all specimens. Based on the measred strength data and the analysis of a free body diagram of a fractred weld, they developed an expression that related weld strength to the loading angle. Later, Lesik and Kennedy (990) formlated a 6

19 simplified version of the strength expression proposed by Miazga and Kennedy (989). The expression of Lesik and Kennedy (990) takes the following form:.5 θ = 0 ( sin θ) (2.) where θ is the load capacity of the fillet weld sbjected to any direction of loading, 0 is the load capacity of a longitdinal fillet weld of same size, and θ is the angle between the line of action of the load and the axis of the fillet weld. Eqation 2. is commonly sed in North American design standards to accont for the effect of the angle of loading on fillet weld capacity. Bowman and Qinn (994) condcted an experimental investigation of geometrical factors that inflence the behavior of fillet welds. The experimental program had 8 specimens, inclding longitdinal and transverse weld specimens with three leg sizes, namely, 6.35 mm (/4 in.), 9.53 mm (3/8 in.), and 2.7 mm (/2 in.), and three different root gap configrations. The welds were prepared sing E708 electrodes. The ratio of the strength of transverse welds to that of longitdinal welds ranged from.3 to.7 for the n-gapped specimens and.2 to.4 for the gapped specimens. Ng et al. (2002) condcted tests on 02 transverse fillet weld connection specimens prepared primarily sing the flx core arc welding (FCAW) process, althogh nine specimens were prepared sing the shielded metal arc welding (SMAW) process to provide a direct comparison with the test reslts from Miazga and Kennedy (989). Of 02 specimens, 96 were doble lapped joints and six were crciform connections. All specimens were concentrically loaded in tension and deformations were measred across the weld leg sing LVDTs. The tests were designed to investigate the effect of the following parameters: () filler metal classification, both with and withot a toghness reqirement; (2) welding process, flx cored vs. shielded metal arc melding; (3) weld size and nmber of passes; (4) welding electrode manfactrers; (5) steel fabricators; (6) low temperatre; and (7) root notch orientation of fillet weld, i.e. lapped vs. crciform splice. The capacity was predicted sing Eqation (2.), since it has been adopted by both CSA- S6-0 (CSA 200) and the AISC Specification (AISC 2005), and the measred weld material strength and fillet weld size. The ratio of tested to predicted strength ranged from.4 to A reliability analysis was performed for different grops of 7

20 connections and the safety index, β, was fond to be at least eqal to 4.5, which is the traditional target vale for connections in the Canadian design standard. Deng et al. (2003) presented the reslts of 8 tests on joints with longitdinal and 45 fillet welds that were tested to complement the test program reported by Ng et al. (2002). The test specimens were prepared with 2.7 mm fillet welds and three different FCAW electrode classifications. It was reported that the strength of fillet welds increased with increasing loading angle. It was fond that the design eqation crrently sed in North America provides an adeqate level of safety for joints with single orientation fillet welds. 2.3 Strength of Crciform Fillet Weld Specimens ham (983a) reported reslts of 36 tests on crciform specimens, of which 8 were made with the FCAW process and 8 were made with the SAW process. The nominal leg sizes were 6 mm, 0 mm and 6 mm. The primary objectives of the tests were to assess the effects of leg size on the strength of fillet welds and to obtain a corresponding load deformation relationship. ham (983a) did not provide reslts of transverse welds in lapped joints. The reslts of his test program, presented in Table E9.5 of Appendix E, show that the ratio of measred strength to predicted strength sing Eqation (2.) is.05 for both FCAW and SAW specimens. It is postlated by Miazga and Kennedy that the prediction models proposed by Miazga and Kennedy (986, 989) might apply to doble fillet T-joints in tension. The predicted fractre for sch joints is 22.5º to the axis of the stem of the Tee and the weld strength wold be at least.34 times the longitdinal vale. Ng et al. (2002) tested six crciform specimens made with E70T-4 and E70T7-K2 electrodes. The test reslts showed that transverse fillet welds in a crciform configration tend to provide both lower weld strength and lower dctility than transverse welds in a lapped splice connection. 8

21 2.4 Load of Joints with Mltiple Orientation Fillet Welds (MOFW) Ligtenberg (968) reported 223 MOFW connections loaded concentrically and 54 specimens loaded with ot-of-plane eccentricity. The reslts showed that the ratio of the area of transverse welds to that of longitdinal welds was the most critical factor that affected the strength of an MOFW connection. A statistical analysis of the test reslts indicated that less than 00 percent of the capacity of the transverse and longitdinal welds was developed in a MOFW. Based on the observation that longitdinal welds have more dctility than transverse welds, Kato and Morita (969) proposed a participation factor for the contribtion of the longitdinal welds to the capacity of a MOFW joint. The contribtion factor, μ, which represents the fraction of the longitdinal weld strength that is developed in a MOFW joint, is given as: μ= /z, 0.28 z 3.6 (2.2) where z is the ratio of leg size of longitdinal welds to that of transverse welds in the MOFW joint. Eqation (2.2) was sed to predict the strength of the test specimens from the international test series of Ligtenberg (968). The ratio of the tested to predicted capacity was.007 for St.37 steel specimens and.023 for St.52 steel specimens. Callele et al. (2005) expanded the research program of Ng et al. (2002) and Deng et al. (2003) to inclde MOFW joints to investigate whether the least dctile segment in a concentrically loaded MOFW connection can deform sfficiently to develop the fll strength of the more dctile segments. A total of 3 doble lapped joints that combined transverse welds with either longitdinal or 45º welds were tested. Complementary tests, inclding nine longitdinal and three transverse fillet welds, were condcted to define the load deformation crves for fillet welds. The parameters considered in the MOFW specimen design were: () fillet weld leg size, (2) nmber of weld passes, (3) weld continity at the corners, (4) weld length, and (5) stress state of the connection plates, which is characterized by the plates remaining elastic throghot the test or the plates yielding. Callele et al. (2005) proposed the following eqation to calclate the capacity of the varios segments of a MOFW: 9

22 Segment = θ CRF(θ) (2.3) where θ is the capacity of the weld segment predicted by eqation (2.), CRF(θ) is the redction factor, which acconts for the fact that the segment strength may not be flly mobilized. For a MOFW joint with combined longitdinal (θ=0º) and transverse (θ=90º) fillet welds, CRF(0º) = 0.85 and CRF(90º) =.0. The capacities of tested MOFW specimens were predicted sing eqation (2.3) and the test-to-predicted ratio was However, a reliability analysis showed the safety index is acceptable for a resistance factor of 0.67, as specified in CSA-S Effects of Weld Size and Length on Fillet Weld Strength Ligtenberg (968) examined the effect of weld size on the strength of connections with transverse welds only and longitdinal welds only. Within the limits of the weld throat dimensions tested, i.e., for weld throats between 3 and 0 mm, with a rather small weld length, the throat size had no significant effect on the strength. However, similar tests by Bornscheer and Feder (966) on 35 specimens welded with rtile electrodes showed that the weld size effect on weld strength was significant. The nit strength of 2 mm longitdinal fillet welds was abot 35% smaller than that of 4 mm welds. The strength of 2 mm transverse fillet welds was abot 20% lower than that of 4 mm welds. Reslts of Bornscheer and Feder (966) also demonstrated that the length of longitdinal fillet welds (length-to-size ratio of 25, 50, 75, and lengths varying from 00 mm to 300 mm) had no inflence on the strength. Kato and Morita (969) performed tests similar to those presented by Ligtenberg (968). In their tests, the weld leg size ranged from 5 mm to 40 mm for transverse welds and from 5 mm to 22 mm for longitdinal welds. The strength was calclated on the measred failre area. Kato and Morita conclded that the average maximm strength of fillet welds was not affected by the size of weld within the large range of leg sizes examined. 0

23 Higgins and reece (969) investigated the effect of weld length pon weld strength. The weld leg size was 6.35 mm (/4 in.) and the length varied from 6 to 6 times the weld size. Test reslts showed the strength increased slightly with the increasing length. ham (98) investigated the effect of weld size pon the weld strength by testing welds with in-plane load eccentricity. The nominal weld leg sizes were 5 mm, 0 mm and 6 mm. The other factors considered in specimen design were the weld length and magnitde of eccentricity. ham (98) fond that the nit strength of small welds was larger than that of large welds. ham (983a) investigated the effect of weld size on the strength of fillet welds sing IIW-recommended crciform specimens for FCAW and SAW processes. The reslts showed that the size effect was most dominant in FCAW for welds p to 8 mm measred at the throat. For larger welds there was no frther redction. For SAW specimens the size effect was more gradal bt covered the whole range of weld sizes investigated in the test series. In a separate investigation, ham (983b) confirmed the observations of the earlier investigation sing Werner specimens in which the welds are loaded longitdinally. Bowman and Qinn (994) tested specimens with longitdinal and transverse welds with leg sizes of 6.35 mm (/4 in.), 9.53 mm (3/8 in.) and 2.7 mm (/2 in.) to investigate the effect of weld size on weld strength. The strength was calclated on the measred rptre areas. For longitdinal specimens, the average strength of 2.7 mm welds was 25% lower than that of 6.35 mm welds. For transverse specimens, the average strength of 2.7 mm specimens was only 4% lower than that of 6.35 mm specimens. It was conclded that a dimple in the exposed weld profile of the 2.7 mm welds de to mltiple passes was responsible for the redced strength. Ng et al. (2002) reported tests of transverse welds prepared with for different FCAW electrodes. The average ratio of the nit strength of the 6.35 mm welds to 2.7 mm welds was.26.

24 Callele et al. (2005) analyzed the effect of weld size and nmber of passes on weld strength sing data from Ng et al. (2002), Miazga and Kennedy (989) and their own tests. Generally, the small size welds were fond to have higher nit strength with larger scatter. The nmber of weld passes seemed to have no effect on strength. Callele et al. (2005) also analyzed the effect of longitdinal weld length on weld strength sing data from Deng et al. (2003), Miazga and Kennedy (989) and their own tests. The lengths of fillet welds examined were 50, 80, 00 and 50 mm. The 50 mm welds had the highest average strength and the 00 mm welds had lowest strength, which showed the decrease in strength with an increase of length. However, the 50 mm welds showed higher strength than the 80 mm and 00 mm welds, which made the investigation inconclsive. 2.6 Smmary Considerable experimental research has been performed on lapped splice connections with fillet welds loaded at varios angles to the weld axis. Researchers generally agree that the weld strength increases as the weld orientation changes from longitdinal to transverse. A reliability analysis of test data from the University of Alberta has shown that the crrent North American design eqations provide an acceptable safety index. The strength of fillet welds in crciform joints has not been given mch attention yet. The reslts from a limited nmber of tests have shown that the strength of crciform connections tends to be lower than that of transverse welds in lapped splice joints. It is qestionable whether the crrent design eqation for weld strength yields an acceptable level of safety when sed with crciform joints. When fillet welds of different orientations are sed in a MOFW connection, the capacity of the more dctile segment cannot be developed flly becase of the limited dctility of the least dctile segment. Callele et al. (2005) have proposed an eqation to consider both the redction of strength of more dctile weld segments and the increase of strength of the less dctile weld segments compared with that of longitdinal welds. A reliability analysis demonstrated an acceptable level of safety based on limited test reslts. The 2

25 applicability of the formla needs to be examined frther with a larger nmber of test reslts. Some test reslts have shown that smaller size fillet welds tend to provide higher nit strengths, no matter whether the strength is calclated on the nominal throat area or on the measred fractre area. Therefore, reliability analysis reslts based on data from tests on small weld sizes (sch as /4 in.) shold be sed with cation. No effect of weld length on weld strength was generally observed. It is therefore conclded that frther experimental research on the strength of crciform connections is reqired and a reliability analysis on SOFW and MOFW joints based on a larger data pool is desirable. 3

26 CHATER 3 EXERIMENTAL ROGRAM AND RESULTS 3. Introdction A review of the literatre on fillet weld research showed that most experimental programs have sed doble lap joints, with limited tests on crciform connections. Becase of the more severe root notch orientation in crciform specimens, it is possible that the strength and dctility of fillet welds in joints of that configration might be significantly lower than the strength and dctility of doble lapped joints. Therefore, an experimental program was designed to investigate the strength and dctility of fillet welds in crciform connections in order to expand the database of test reslts on concentrically loaded fillet weld connections. This represents the forth phase of an ongoing research program on welded joints condcted at the University of Alberta. hase inclded tests on transverse welds condcted by Ng et al. (2002), phase 2 refers to the test program by Deng et al. (2003) and phase 3 refers to the work of Callele et al. (2005). The tests presented in this chapter were condcted by Mr. Andre Blanchard and Mr. Derek Kramar, nder the spervision of Mr. Logan Callele. 3.2 Test arameters The factors considered in the design of the test matrix are: connection configration (i.e. crciform connection vs. lapped splice specimens from previos phases of the research program), welding electrode classification, fillet weld size, main plate stress condition, and test temperatre. The phase 4 test matrix, inclding for different test cases, is presented in Table 3.. The test specimens are designated by the letters CNY, indicating a crciform joint configration with "non-yielding" plates, i.e., plates designed to remain elastic throghot the tests. The specimens are nmbered seqentially from to 2. Each test was repeated three times to assess variability of the test reslts within one treatment. The specimens were prepared sing the flx-cored arc welding (FCAW) process. Two FCAW filler metals were selected, namely, an E70T-7 filler metal, which has no 4

27 toghness reqirement (AWS, 995), and an E7T8-K6 filler metal, which has a toghness reqirement of 20 J at 29 ºC (AWS, 998). The nominal tensile strength for both electrodes is 480 Ma (70 ksi) and the actal tensile strength was established sing tests on all-weld-metal tension copons. Three sch copons, with a 50 mm gage length, were prepared in accordance with the appropriate AWS specification (AWS, 995, 998) from each wire spool. The reslts of six copon tests are presented in Table 3.2 and the stress verss strain crves from each tension copon are presented in Appendix A. The steel plates sed for the fabrication of the test specimens conformed to the reqirements of ASTM A572 Grade 50 and CSA G W. This grade of steel is sitable for welding bt has no toghness reqirement. All specimens were fabricated with plates from the same heat. The test reslts from phases to 3 indicated that weld size has a significant effect on the nit strength of fillet welds. Test specimens with a weld size of 6.4 mm showed a higher nit strength than 2.7 mm welds. Since the six pilot crciform specimens in phase had a leg size of 6.4 mm, all the test specimens from the crrent phase were made with (nominally) 2.7 mm welds. Another factor considered in specimen design is the average stress level in the plates dring testing. Test reslts from Ligtenberg (968) indicated that stress level in the plates did not have a significant effect on fillet weld strength. Callele et al. (2005) examined the test reslts from phases and 3 and those of Miazga and Kennedy (989) and fond that plate yielding before weld fractre may increase the weld dctility, bt has negligible effect on weld strength. The plates sed in this phase of the test program were therefore designed to remain elastic. The effect of low temperatre on fillet weld behavior was also one of the test parameters for this phase of the test program. Six crciform specimens were tested at -50 C. Unfortnately the specimens did not fail as expected. All the specimens tested at low temperatre fractred in the plates. Therefore, the low temperatre test reslts are not inclded in the analysis presented in Chapters 3 and 4. A discssion of the low temperatre tests can be fond in Appendix D. 5

28 3. 3 Specimens Description The dimensions of the test specimens are shown in Figre 3.. As depicted in the figre, three 76 mm crciform test specimens were obtained from each assembly. The assemblies were fabricated so that both sides of the connections were nominally identical. Once the assemblies were fabricated they were inspected visally for weld qality and conformance to the design specifications. Whichever side of the central plate was determined to have better weld qality was taken as the test welds and the other side was reinforced by adding additional weld passes in order to force the failre in the test welds, ths redcing the nmber of welds that had to be instrmented dring testing. The test specimens were prepared from the weld assemblies by water jet ctting, followed by milling. 3.4 re-test Measrements For types of measrements were made to characterize the fillet weld before testing: main plate leg size (ML), central plate leg size (CL), throat dimension (measred at 45º), and length of the two test fillet welds. A description of the weld dimensions measred is shown in Figre 3.2. The ML and CL dimensions and the 45º throat dimension were measred at eight locations spaced eqally along the two test welds for each test specimen. The devices sed for measrement of weld size were described in detail in Callele et al. (2005). A smmary of the measred weld sizes is presented in Table 3.3, where the vales shown represent the average of eight measrements. The complete set of measrements is reported in Appendix B. For crciform specimens, any misalignment of the main plates reslts in load eccentricity dring testing. The eccentricities of the main plates were measred as depicted in Figre 3.3 and the reslts are presented in Table 3.4. The reslts show that the maximm eccentricity was only 2.05% of the thickness of main plate. 6

29 3.5 Instrmentation and Test rocedre The crciform specimens were tested in a MTS 6000 niversal testing machine and loaded in concentric tension ntil rptre of the fillet welds or plates occrred. The specimens were oriented so that the weld axis was horizontal and the test welds were positioned below the reinforced welds. This orientation facilitated the instrmentation of the specimens with linear variable differential transformers (LVDTs). The overall test setp and instrmentation for the specimens are depicted in Figre 3.4. Special LVDT brackets were designed in phase to measre the deformation of the test welds dring loading. The LVDTs were monted on the test specimens as shown in Figre 3.4. Two hardened steel anchors sed to spport one end of the monting bracket were set in two light pnch marks made on the base plates at the toe of the welds. These pnch marks ensred that the anchors of the brackets remained in place dring the test. The rear of each bracket was fitted with two rollers to stabilize the assembly while at the same time eliminating longitdinal restraint. The brackets were kept anchored to the pnch marks dring the test sing rbber bands wrapped arond the specimens. The gage length over which the weld deformations were measred was the distance between the pnch marks and the face of the central plate. The pnch marks were placed as close as possible to the toe of the weld so that the amont of plate deformation captred within the gage length was kept to a minimm. The gage length sed for strain calclations was taken as the average weld leg size within the instrmented weld segment. For specimens CNY-6, 7, 8, 0,, 2, electrical resistance strain gages were sed to measre the strains at the edges of the test specimens, near the weld toe. The strain gage arrangement is depicted in Figre 3.5. These strain gages were sed to assess the amont of load eccentricity both in the in-plane and in the ot-of-plane directions. The specimens were loaded qasi-statically nder displacement control. The load and displacements were recorded in real time dring the tests. Static load vales were acqired by maintaining a constant deformation for abot five mintes. The tests were terminated when one of the test welds rptred. The instrmentation was then removed 7

30 and the specimen was loaded ntil both test welds had rptred so that the weld fractre srfaces cold be inspected. 3.6 Main lates Misalignment Second Order Effects Althogh precations were taken, fabrication imperfections in the test specimens were navoidable. Both an anglar misalignment and an offset of the axis of the main plates (see measrement and 2 in Figre 3.3 for example) case bending of the main plates and affect the stresses in the welds dring the test. Neglecting the small offsets, the relationship between the maximm bending stress in the plate, σ, reslting from the second order effects, and the tensile stress, σ t, cased by the axial force can be estimated by the following eqation if the plates are known to remain elastic and the test machine grips are considered pinned: 6e σ σ = (3.) h b t where e is the eccentricity of the applied load at the section of interest and h is the dimension of the cross-section perpendiclar to the bending axis, as shown in Figre 3.6. The vale of σ / σ for the six specimens tested at room temperatre was analyzed and b t the reslts are smmarized in Table 3.5. The load eccentricity sed in the calclations is the maximm of the for measrements for each specimen as presented in Table 3.4 (i.e., the maximm of L, L2, L5, and L6). A positive eccentricity cases an increase in tensile stress on the front face of the test specimen as shown in Figre 3.6. The section height (main plate thickness), h, was 50.8 mm. Table 3.5 indicates that the bending stress on the srface of main plates cold be as high as % of the tensile stress dring the tests. This means that one of the two welds wold potentially carry more load than the other weld. However, this observation is based on the assmption that the material remains elastic throghot the test, which was not the case. Yielding of a weld segment wold case a redistribtion of the load carried by each weld segment, ths redcing the eccentricity effect. In fact, the fractred welds were not always at the face with higher tensile strains, bt were always at the face having the least b 8

31 weld throat area. This reslt indicates that in reality the bending effect cased by the eccentricities was not as large as is implied by Table 3.5. The strain measrements from the six specimens tested at room temperatre are presented in Appendix C. Figres C4 to C9 show plots of the normalized load, fnction of the normalized strain, ε / ε y. /, as a Figres C20 to C25 present plots of the strain ratio, ε / ε as a fnction of the load ratio, /. Generally, the bending strains reached a peak at abot less than 0% of the ltimate load and then decreased sharply to a vale close to zero. The stress concentration cased by the abrpt change in geometry made the strains measred with strain gages very different from the average strain in the plates and it is likely that localized yielding near the weld toe occrred early casing the forces in the welds to redistribte. Overall, the fabrication imperfections are not considered to have had a significant effect on the behavior of the specimens. 3.7 Crciform Specimen Capacities The ltimate static capacity of each specimen tested at room temperatre is presented in Table 3.6. The strength of the fillet weld was obtained by dividing the static ltimate load ST by the throat area of the specimens as measred before testing. The minimm throat area A throat of each segment is a fnction of the measred weld leg size on the main plate, ML, and on the central plate, CL, and the weld length. The definition of ML, CL and minimm throat dimension (MTD) for a typical fillet weld cross-section is illstrated in Figre 3.2. The minimm throat dimension (MTD) is the shortest distance from the root of the fillet weld to the hypotense of the triangle defined by the two legs of the weld. The minimm throat dimension is obtained from: b t MTD = ML CL ML + CL 2 2 (3.2) 9

32 The minimm throat area, A throat, is therefore eqal to the prodct of the minimm throat dimension, MTD, and the weld length. The throat area ths obtained ignores the weld root penetration and the weld face reinforcement, if any. The total throat area is calclated differently depending on the nmber of fractred welds. If only one weld fractred, the total throat area is taken as twice the minimm throat area of the fractred weld. This, in effect, assmes that half of the applied load was carried by the fractred weld. If both welds fractred simltaneosly, the throat area is taken as the sm of the minimm throat areas of the two welds. The specimens were loaded qasi-statically nder displacement control. The load and displacements were recorded in real time dring the tests. Static load vales were acqired by maintaining a constant deformation for abot five mintes and the pper load is the extreme large vale and the lower load is the extreme small vale within the five minte maintenance of the deformation. The static drop is the difference between the pper and lower load. The ltimate load is the extreme large vale of load dring the entire loading process. The ltimate load may occr within the last five minte maintenance of deformation or after. The static load reported in Table 3.6 is the one recorded near the ltimate load. 3.8 Weld Strains Calclated from Deformations Measred with LVDTs The weld segment deformations and the calclated strains at the ltimate load and at fractre of specimens tested at room temperatre are presented in Table 3.7. The weld strains in the direction of applied load are obtained by dividing the measred weld deformation, Δ, by the average main plate leg (ML) dimension, The measred deformation Δ is reported differently depending on the nmber of fractred welds. If only one weld fractred, the deformation is taken as the average deformation from the two LVDTs monted on the fractred weld. Otherwise, the deformation is taken as the average of the for LVDT measrements on the two test welds. The response crves for the specimens tested at room temperatre are presented in Appendix C. * d. 20

33 3.9 Effect of Root Notch Orientation on Weld Strength The normalized ST A throat vales presented in Table 3.8 were obtained by dividing A by the measred weld metal tensile strength for the corresponding spool of ST throat filler metal. The normalized strength of crciform specimens is compared with that of lapped splice specimens with transverse welds from phases and 3. These reslts are tablated in Tables 3.8a throgh 3.8d and presented graphically in Figres 3.7a to 3.7d. Each of the figres indicates both the range of vales and the mean for a particlar set of variables. Table 3.8a compares the normalized strengths for the lapped splice joint configration to those obtained from crciform specimens with a leg size of 2.7 mm and fabricated sing E70T-7 welding electrodes. The normalized strengths are plotted in Figre 3.7a. The ratio of the mean normalized strength for crciform joints to that of lapped splice specimens is 0.8. Table 3.8b and Figre 3.7b present a similar comparison for test specimens with a leg size of 2.7 mm and E7T8-K6 filler metal. The ratio of the mean normalized strength for crciform joints to that of lapped splice specimens is The reslts indicate that the strength of crciform joints is significantly lower than that of eqivalent lapped splice joints. Table 3.8c presents a comparison between the normalized strength for lapped splice joints and crciform joints for 6.4 mm welds and E70T-4 welding electrode classification from phase of the University of Alberta weld test program. The range of the normalized strength is depicted in Figre 3.7c. The ratio of the mean normalized strength for crciform joints to that of lapped splice joint specimens is.0. A similar comparison between crciform joints and lapped splice joints made with 6.4 mm welds of E70T7-K2 filler metal is presented in Table 3.8d and Figre 3.7d. In this case, the ratio of the mean normalized strength of crciform joints to that of lapped splice joints is The reslts indicate that the strength of crciform joints is eqal to or lower than that of lapped splice joints with 6.4 mm weld size. The strength ratios presented in Tables 3.8a to 3.8d are smmarized in Table 3.9. The comparisons indicate that the root notch orientation in the crciform specimen reslts in 2