DEVELOMENT OF DUCTILE SEMI-RIGID JOINTS WITH LAGSCREWBOLTS AND GLUED-IN RODS. Yoshiaki Wakashima 1, Kenho Okra, Kazo Kyotani ABSTRACT: Althogh the dctility of joints is important from the viewpoint of the seismic response, joints with lagscrewbolts or gled-in rods normally exhibit very stiff elastic behavior that cases brittle failre of wood. The prpose of this stdy is to obtain dctile semi-rigid portal frame joints sing lagscrewbolts or gled-in rods sch that brittle failre of the joints is prevented. We developed two types of leg and beam-colmn joints. With regard to dctility, the performance of the joints is governed by the yield and plastic deformation of steel. The diameters of connecting bolts, which penetrate the lagscrewbolts, are chosen sch that the connecting bolts yield nder the shear strength of the joint panel. Similarly, for joints sing gled-in rods, threaded steel rods have a redced cross-section in a certain length so as to yield the thinner part. Cyclic loading tests were condcted for the developed joints and for portal frames having these joints. A large deformation was observed with little damage of the wood members. Analytical models of the joints were examined considering prestress in the bolts. An analytical model of the portal frames was also examined considering the shear deformation of the joint panel and bending property of the colmn. The calclated reslts were in good agreement with the experimental reslts. KEYWORDS: ortal frame, Semi-rigid joint,, Gled-in Rod, Dctility 1 INTRODUCTION 1 Axial fasteners sch as gled-in rods and lagscrewbolts normally exhibit very stiff elastic behavior nder tensile forces withot any particlar consideration [1]. Therefore, joints with the fasteners parallel to the grain, sch as leg joints, are very brittle. Similarly, when the fasteners are embedded perpendiclar to the grain, sch as in a beam colmn joint, the joints generally trn ot to be brittle de to shear failre in the joint panel. For the above reasons, in order to obtain a dctile joint, which is an important reqirement from the viewpoint of the seismic response, it is necessary to ensre that the tensile strength of the fastener is controlled so that brittle failre is prevented. One of the athors has investigated joints composed of gllam frames and special steel connectors, where gled-in rods attach the connectors to the frame []. The yield strength of the steel connectors was designed to be lower than the strengths of the gllam frames and gled-in rods so that most of the damage wold be concentrated within the steel connectors. These joints showed reasonable dctility and energy 1 Yoshiaki Wakashima, Toyama refectral Forest rodcts Research Institte, 494 Krokawa-shin, Imiz 99-11 Japan. Email: yoshiaki.wakashima@pref.toyama.lg.jp Kenho Okra, GrandWorks Corporation, 4 Oenoki, Namerikawa 96-874 Japan. Email: k- okra@grandworks.co.jp Kazo Kyotani, Laminate-lab Corporation, 1 Ksajimaazafrkawa, Toyama 9-1 Japan. Email: kyotani@laminate-lab.jp absorption; however, since the connector was completely exposed to the exterior, an architectral design problem arose. In this stdy, we examined joints, the dctility of which is governed by the yield strength of the bolts that are inserted in timber so as to minimize the exposre of the steel materials to the exterior. JOINTS.1 STRUCTURE OF NEW JOINT SYSTEMS We developed two types of leg joints shown in Figre 1. RC-type joints consist of gled-in rods that have redced cross-sections withot ribs in a certain length so that the rods yield. The material of the rods is mild steel. Nts are also installed between the colmn and the steel base so that the rods resist compressive forces. s are sed for LC-type joints. Their performance is governed by the yield strength and plastic deformation of the anchor bolts. Two types of beam-colmn joints are developed as shown in Figre. In C-type joints, the bolts, which pass throgh the central hole of the lagscrewbolts embedded in the colmn, are connected to the lagscrewbolts embedded in the beam. In B-type joints, a hole having a particlar length is pnched in the lagscrewbolts embedded in the beam, and the bolts are connected from a steel plate to the inner hole of the lagscrewbolts. A detailed illstration of the plates is shown in Figre. For both the beam-colmn joints, the diameter of the connecting bolts is decided sch that the bolts yield
nder the shear strength of the joint panel. The anchor bolts and the connecting bolts for beam-colmn joints described as ABR49 in Figres 1 and are roll threaded bolts. The effective sectional area of the bolts is almost same between body and thread, which lead to yielding of bolts niformly and large plastic deformation. Nominal diameter of the lagscrewbolts sing above joints is mm. ltimate moment exceeded the reference strength of the gllam. Becase of the reaction of the nts to compressive forces, little slip behavior was observed in the hysteresis loop. Tensile failre of the anchor bolts occrred in the LCtype joint; we obtained a large rotation angle exceeding.1 rad. However, nlike RC-type joints, slip behavior is observed in the hysteresis loop becase the anchor bolts. JOINT TESTS Static cyclic loading tests were performed for the developed joints. Figres 4 and illstrate the details of the test specimens and set-ps. Table 1 lists the specifications of the investigated test specimens. The nominal diameters of the anchor bolts are 14 mm and 16 mm for LR and LR4, respectively. The effective elongations of the bolts (shown in Figre ) are 79. mm and 19 mm for CL and BL4-type joints, respectively. The relationships between moment and joint rotation for leg joints are illstrated in Figre 6. Since the failre of the RC-type joint was fractre of the rods, reasonable plastic behavior was observed. The observed Gle-in rod Redced diameter Nts for reacting to compression force Fondation (a)rc-type (b)lc-type Figre 1: Developed leg joints. JSS: Japan Steel Standard Shear Key Steel Base late Anchor Bolts JSS ABR49 Sill Effective elongation length of bolt Bolt JSS ABR49 Lag Bolt Steel Gsset Shear Key Steel late Bolt JSS ABR49 Effective elongation length of bolt Steel late Shear Key Drift-pin Drift-pin (a)c-type Figre : Developed beam colmn joints (b)b-type Figre : Steel plates for B- type joints Actator GIR(M) JIS SS4 φ1 7 Nt 18 Steel colmn base (a)rc-type (b)lc-type Figre 4: Test specimens for leg joints. JIS: Japan Indstrial Standard 1,419 Elongation Length of Ancor Bolt =8mm Shear Key Steel Base Anchor Bolt JSS ABR49 Fondation Jig (H-Shaped Steel) in spport bar 1 Figre : Test set-p for beam colmn joints Table 1:Specification of joint specimens Code Cross section Wood material JAS grade name (mm) RC Gllam(Sgi) E6-F 18 4 Leg joint LR Gllam(Red pine) E1-F 1 LR4 Gllam(Red pine) E1-F 1 4 CL Gllam(Red pine) E1-F 1 colmn joint BL4 Gllam(Red pine) E1-F 1 4 16
1 1 - -1-1 - -.1.1...4. (a)rc-type (b)lr-type (c)lr4-type Figre 6: Moment- rotation relationships for leg joint specimens do not resist compressive forces. The rotational rigidity of the leg joints is calclated sing the analytical model proposed by Koizmi et al []. In order to determine the maximm strength of the joints, a netral axis, which varies according to yield conditions of the bolts, shold be decided. When the rods on the tensile side are in the yield condition as shown in Figre 7(a), the netral axis is expressed in the form of eqation (1) 4 1-1 - - -4 -...1.1.. γ Kd δ 8 6 4 Reaction forces h1 - -4-6 -...1.1. g δ g1 l b Q M Reaction forces d t d c 1 = (4δ + 4Kdδ + b eh1 Kw( e+ δ ) be Kw( e+ δ ) (a)rc-type (b)lr-type Figre 7: Yield condition of leg joint at the maximm strength ± ( 4( b e h1 Kw( e+ δ ) + δ ( + Kdδ )) 4b e Kw( e+ δ )( be h1 Kw+ δ (h1 + ( g1+ g) Kdδ ))) (1) where γ is length of the yield section of wood, Kd is the slip modls of the gled-in rod, δ is the slip of gled-in rod, δ is the ltimate slip of the gled-in rod, is the yield strength of the gled-in rod, Kw is the embedded rigidity of wood, e is yield embedded displacement of wood, and b is the width of the colmn. The above eqation is derived assming that the embedment resistance of wood and the tensile resistance of the rods are perfectly elasto-plastic. The calclated reslts are shown in Figre 6(a) as dotted lines. The experimental and theoretical reslts were in good agreement. As mentioned above, a large ltimate moment was observed. This may be becase of two reasons. One reason is that the yield stress in the joints is distribted as shown in Figre 7(a), which leads to the stress being restrained in the oter laminations of gllam. Another 1 1 - -1-1 Experiment Calclate - -...1.1. - -. -. -.1.1...7.9.11 (a)cl-type (b)bl4-type Figre 8: Moment- rotation relationships for beam colmn joint specimens 4 1-1 - - -4 probable reason is that the longitdinal stress of wood near the fixed end of the colmn is not sfficiently large so that little failre occrs de to bending. This assmption is mentioned in following section. Since plastic deformation of the anchor bolts cases deformation of the leg joint, eqation () is applied to estimate the rotational rigidity of LC-type joints [4]. Es nt Ab Kbs= α l ( d + d ) t B b where d t, d c, and l b are as illstrated in Figre 7(b). Es is Yong s modls of the anchor bolt, A B is the crosssectional area of the anchor bolt, and n t is the nmber of anchor bolts on the tensile side. The vale of α B is generally for bending deformation of the base plate and deformation of concrete on the compression side [4]. However, α B is ignored in this stdy becase the base plate has sfficient thickness and becase H-shaped steel jig is sed as a sbstitte for the concrete fondation. The calclated reslts and the experimental reslts are in good agreement as shown in Figre 6(b) and 6(c). Since the bending of the base plate and deformation of the fondation jig are ignored, it seems that frther verification is necessary to simlate the actal sitation by setting a leg joint on concrete. The experimental reslts for beamcolmn joints are illstrated in Experiment Calclate c () Figre 8. From the reslts, a large deformation cold be obtained by plastic deformation of the bolts withot any visible failre of the gllam. Slip behavior is not clearly observable in the hysteresis loop for L-type joints. The reason for this reslt is that the bolts that are
connected to the lagscrewbolts embedded in the beam are considered to resist both tensile and compressive forces, becase the bolts behave as bckling-restrained members in the lagscrewbolts embedded in the colmn. It is considered that the rotational rigidity of beamcolmn joints sbjected to open-mode moment is different from that of joints sbjected to close-mode moment. Figre 9 shows the mechanical models for joints sbjected to open- and close-mode moment. From the eqilibrim of moments, eqation () and (4) are obtained for the opening and closing modes, respectively, by assming a beam to be a rigid body in the section of the joint. ( g) ( h ) La+ h + + Kd1 h g) Kd La () b M= Kw ( g) ( h ) La Kd c( La g h) b M = Kw La g+ + Kd1 δ + where is the netral axis, b is the width of the colmn, Kw is the embedment rigidity of wood, Kd1 is the tensile slip modls of the fastener, Kd is the compressive slip modls of the fastener, and is the embedded displacement at the beam end. The above eqation can be expressed in the form of a cbic eqation as a fnction of when b, c, and d are appropriately introdced. (4) d + c+ b + = () Then, eqation () can be solved sing the soltions for cbic eqations. For example, the following notations can be introdced. b b bc 4 p = c q = + d r = p 7 q p D= + If D >, then π 1 q b = rcos( ArcCos[ ]) (6) p r The compressive force C, which is prodced de to the compression of the bolts and the embedment of wood, is expressed as 1 Kd1( g) C = b Kw+ (7) The tensile bolt force T is Joint panel La Joint panel (a)opening mode (b)closing mode Figre 9: Mechanical model for beam-colmn joins La Elastic Fondation Kd ( g ) T = (8) From eqilibrim condition C + T + = = g Kd1 h Kd+ The joint moment is ( Kd1+ Kd + b Kw) h ) = 1 (9) Mj (1) From eqations (6), (9), and (1), the rotational rigidity of joints is obtained as Mj Kj= ( δ c / ) (11) When F is a shear force that occrs at the joint panel as shown in Figre 9, C is eqal F for the opening mode and T is eqal F for the closing mode. The shear strain γ prodced by F is γ = α F /( Gw Aw) (1) where Gw is the shear modls of the colmn, Aw is the cross-sectional area of the colmn, and α is the shear coefficient. Eqation (1) is the rotational rigidity of the joint considering shear deformation of the colmn in the joint. Kj = Mj ( δ c / + γ) (1) Becase the effect of bending deformation of the colmn in the joints is ignored in the above eqations, finite element (FE) analysis, in which the colmn is assmed to be a beam on an elastic fondation as illstrated in Figre 1, was carried ot to evalate the effect of bending deformation. Figre 11 shows the calclated displacements at h1 and g1 obtained from eqation (9) and FE analysis. From these reslts, it is clear that the location of the netral axis and the displacements at h1 and g1 are almost coincident and the assmption that the colmn is a rigid body in the section of the joint seems to be appropriate. Since a tightening torqe of 1 Nm is applied to the bolts that are sed to assemble the beam and colmn, it is considered that the effect of prestress in the bolts shold be taken into accont in order to estimate the rigidity of C-type joints. When the beam is embedded in the colmn as shown in Figre 1(a), eqation (1) is obtained from the moment eqilibrim. Gle-in rods Figre 1: on elastic fondation model for beamcolmn joints Displacement(mm)...1 -.1 -. h1 Netral Axis FE analysis Calclate Figre 11: Calclated displacements at h1 and g1 g1
( g) ( h ) 1 La Kd c( La g h) b M = Kw La g+ + Kd δ + b( h+ g ) ( h+ g ) Kw La g+ + (1) When the bolts reach their tensile strength as shown in Figre 1(b), the eqations corresponding to the eqilibrim of the moments for opening and closing modes are expressed by eqation (16) and (17), respectively. ( γ) γ M = Q1 La Q h g) Cw1 La+ h + Cw h ) (16) ( γ) γ M= Q 1 La+ Q h g) Cw1 La g+ γ + Cw ( La g+ ) (17) The initial rigidity of the joints calclated from eqations (), (4), and (1) and the ltimate moment calclated from eqations (16) and (17) are illstrated in Figre 8 as dotted lines. The calclated reslts obtained sing eqation (1), which considers the effect of prestress, are lower than the experimental reslts as shown in Figre 8(a). One possible reason for this reslt is that some measrement error may have occrred when the measrement points were set near the interface between the colmn and beam, since these points are moved following the recover of the embedded displacement of wood that is generated by embedding the beam to the colmn de to the prestress. Althogh the yielding points of the joints in the opening and closing modes differ by 1 1%, the calclated reslts can satisfactorily describe this phenomenon. The difference in the ltimate moments of the opening and closing mode is not significant; however, the shear force prodced in the joint panel at the closing mode is % larger than that at the opening mode. These are important reslts that help designers to appropriately design joints that do not ndergo failre in the joint panel. ORTAL FRAMES.1 ORTAL FRAME TESTS In order to evalate the performance of the developed joints, static cyclic loading tests were carried ot on portal frames sing the developed joints. Table lists the specifications of the test specimens sed in the portal frame tests. Figre 1 shows an example test setp and the actal test setp for the 8F specimen. The specimens are laid on the test floor and colmn bases are fixed on H-shaped steel jigs. The beam was sbjected to psh-and-pll cyclic loading and A i distribtion was applied to calclate the shear force of second story []. A i is a vertical distribtion coefficient of seismic story shear coefficients. 1 T Ai = 1+ (1+ αi ) 1+ T α (18) i Torqe La Colm δy (a)considering torqe (b)final failre inflence condition by fractre of rods Figre 1: Mechanical models for beam-colmn joints Table : Specification of portal frame specimens Code name Span Height Red pine gllam colmn joint Leg joint (mm) (mm) (mm) 1 x 1F 7 1F:777 LR CL JAS-grade:E1-F 81F 78 1F:777 1 x LR4 BL4 F:7 JAS-grade:E1-F 1 x F 7 1F:777 LR CL JAS-grade:E1-F 8F 78 1F:777 F:7 1 x JAS-grade:E1-F LR4 BL4 7 Fondation jig (H-shaped steel) 777 81 696 :7 8:78 (a)schematic diagram Figre 1: Test set-p for fll-scale portal frame specimen (b)hotograph showing testing featre
Shear Force (kn) 4 1-1 - D FEM(1 st ) D FEM( nd ) 1 st story nd story - -1-1 - 1 1 Shear Deformation (cm) Shear Force (kn) 8 6 4 - -4-6 D FEM(1 st ) D FEM ( nd ) nd story 1 st story -8-1 - 1 1 Shear Deformation (cm) (a)f Figre 14: Load-shear deformation obtained from experimental tests (b)8f Where α i =W i /W, W is weight above grond level, T is fndamental natral period. Figre 14 shows the relation between load and shear deformation obtained from the tests. The shear deformation of F de to plastic deformation of joint bolts was so large that the test terminated by the stroke limit of the actator. Both side of the leg joints became pin condition after splitting of the wood arond the lagscrewbolts for 8F, althogh, a relatively large shear deformation occrred. The relationships between the moment and rotation angle of the joints are illstrated in Figre 1 and 16. The calclated reslts are also plotted. The joint moment is determined from the distribtion of bending strain obtained from strain gages and Yong s modls of gllam measred sing the Timoshenko-Goens- Hearmon (TGH) method [6]. The maximm rotation angle of the leg joint on the left side cold not be obtained, since the displacement transdcer separated from the specimen before the test ended. These reslts indicate that the shapes of the hysteresis loops are almost identical to the corresponding loops obtained from the joints tests. The loops of the beam-colmn joints for 8F show a wider area (indicating the energy absorption ability) as compared to the reslts of the joint test, since the frame specimen was sbjected to a large cyclic shear deformation. The rigidity and strength of the beam-colmn joints are calclated as mentioned previosly; however, moment eqilibrim is calclated considering shear forces that act on both sides of the joint in the first story. With the assmption that joint failre is cased by bolt fractre, the maximm angle of joint rotation is determined from 1 1 - -1-1 - -...1.1. 1-1 - (a)left side joint of nd story - -...1.1. 4 1 (c)left side joint of 1 st story -1 - - -...1.1 the strain data obtained by tensile tests of bolts. The calclated vales for leg joints are also given as stated above. The axial forces on the colmns, which are necessary to estimate the strength of the leg joints, are determined from the measred shear strain vales of the beams. The strain data of the anchor bolts obtained from the frame tests are sed for calclating the maximm rotation angle of the leg joints. The observed and calclated reslts show good agreement. The yield moments differ by 1% 1% between the opening and closing mode for the beam- 1 1 - -1-1 - -. -.1 -.1 -.. 1-1 - (b) Right side joint of nd story - -. -.1 -.1 -.. (d) Right side joint of 1 st story - -...1.1 (e)left side leg joint (f) Right side leg joint Figre 1: Moment-rotation relationships obtained from F. :Experiment, -:Calclate. 4 1-1 -
colmn joints of the second story. Becase of the effects of the axial forces of the colmns, the ltimate moments of the leg joints differed by abot % for F and 1% for 8F between the right and left side. This indicates that a large ratio of span to story height decreases the strength of leg joints. The relation between load and shear deformation in onestory specimens is shown in Figre 17. Large deformations were obtained for both 1F and 81F de to plastic deformation of the joint bolts. The final mode of failre was fractre of the anchor bolts and splitting of wood in the beam-colmn joints. The shear deformation of timber portal frames can be calclated by treating them as semi-rigid portal frame. The initial stiffness inclding the effect of shear deformation of the timber elements is illstrated in Figre 17 as dotted lines. The calclated and experimental reslts are observed to be in good agreement. However, it is considered that the shear strain of the joint panel shold be inclded in the estimation of the deformation of portal frames. Figre 18 shows the comparison between the shear strain of the joint panel and the joint moment, which are obtained from the beam-colmn joint of 1F. Althogh the observed shear strain was relatively small, it is considered that this strain shold not be ignored when calclating the deformation of frames. Ths, the effect of shear strain is considered for estimating the rotational rigidity of beamcolmn joints in this stdy. Another problem in estimating the deformation of portal frames is the bending property of the timber in which lagscrewbolts are embedded parallel to the grain. Shear Force (kn) 4 1-1 - - 4 1-1 - - -4 - -. -. -.1.1...7.9 4 1-1 - - -4 - -.11 -.9 -.7 -. -. -.1.1.. 8 6 4 - -4-6 (a)left side joint of nd story (c)left side joint of 1 st story -4 - -1 1 4 Shear Deformation (cm) -8 -. -. -.1.1...7.9 Figre 19 shows the longitdinal strain distribtion of a leg joint specimen simlated by FE analysis. The reslt shows that the maximm tensile strain of the wood occrs arond the top of the lagscrewbolts, and the strain decreases on approaching the colmn end. In Figre, the colmn deflections obtained from the FE analysis reslts are compared to the vales calclated considering the colmn as a cantilever beam. The FE analysis vales are smaller than the calclated vales. When the part of the timber in which lagscrewbolts are embedded is 4 1-1 - - -4 - -.1 -.8 -.6 -.4 -...4 4 1-1 - - -4 - -. -. -.1.1...7.9 8 6 4 - -4-6 (b) Right side joint of nd story (d) Right side joint of 1 st story -8 -. -. -.1.1...7.9.11 (e)left side leg joint (f) Right side leg joint Figre 16: Moment-rotation relationships obtained from 8F. :Experiment, -:Calclate. -8-1 -1-1 1 Shear Deformation (cm) (a)1f (b)81f Figre 17: Load-shear deformation obtained from experimental tests. -:Experiment, :Calclated as semi-rigid portal frame, -:D FE analysis reslts in consideration of the shear strain of the joint panel and the assmption as a composite beam. Shear Force (kn) 1 8 6 4 - -4-6
assmed to be a composite beam comprising timber and lagscrewbolts, the calclated reslts agree well with the FE analysis reslts, as indicated by the dotted line in Figre. The strain distribtion is not linear between the timber and the lagscrewbolts, which indicates that the above assmption does not represent the actal bending behavior of the colmn. However, this assmption is applied as a simplified method to estimate the deflection in this stdy. Figre 1 shows the D FE analysis model employed above assmptions. Initial stiffness of the portal frames obtained from this model is shown in Figres 14 and 17. The experimental reslts and the D FE analysis reslts are in better agreement than withot the considerations of above effects. 4 CONCLUSION This stdy investigated the development of new semirigid joints having performances that are governed by the yield and plastic deformation of steel fasteners. Large rotational angles were obtained in experimental tests. The calclated rotational rigidity of the joints for both open- and closed-mode moment showed good agreement with the experimental reslts. Cyclic loading tests of portal frames sing the developed joints also proved sfficiently dctility. Calclated initial stiffness of portal frames in consideration of the effects of the joint panel and the bending property of timber showed good agreement with the experimental reslts. REFERENCES [1] Gehri E.: Dctile behavior and grop effect of gled-in steel rods. In: roceeding of 1- Longitdinal strain International RILEM Symposim on Joint in timber Strctre, -4, 1 [] Wakashima Y., Sonoda S., Ishikawa K., Hata M., Okazaki Y., Hasegawa K.: Development of response techniqe for timber frame strctre. In:8th World Conference on Timber Engineering, 449-4, 4 [] Koizmi A., Sasaki T., Jensen J. L., Iijima Y., Komats K.: Moment-resisting roperties of ostto-sill Joints Connected with Hardwood Dowels. Mokzai Gakkaishi, 47(1):14-1, 1. [4] Architectre Institte of Japan: Recommendation for Design of Connections in Steel Strctres. Architectre Institte of Japan, Tokyo, 1. [] Architectre Institte of Japan: Recommendation for Load on Bilding. Architectre Institte of Japan, Tokyo, 4. [6] Kbojjima Y., Yoshihara H., Ohta M., Okano t.: Examination of the Method of Measring the Shear Modls of Wood Based on the Timoshenko Theory of Bending. Mokzai Gakkaishi, 4(1):117-1176, 1996. 1 1 - -1-1 Experiment Calclate - -.4 -...4.6 γ Figre 18: Shear strain of joint panel and joint moment relationships for 1F Figre 19: Longitdinal strain distribtion of leg joint specimen obtained by FE analysis Difrection (mm) 9 8 7 6 4 FE analysis Simple beam Composite beam Gllam element Composite beam element Semi-rigid joint element (Rotational rigidity of leg joint) Semi-rigid joint element (Rotational rigidity of beam colmn joint in cosideration of joint panel) Rigid body element 1 4 6 8 1 1 14 Distance from fixed end (mm) Figre : Comparison of the deflection of beam among FE analysis reslts and calclated reslts Figre 1: D FE analysis model