Jtam-A4.dvi JOURNAL OF THEORETICAL AND APPLIED MECHANICS 51, 4, pp. 891-902, Warsaw 2013 GLOBAL BUCKLING PREVENTION CONDITION OF ALL-STEEL BUCKLING RESTRAINED BRACES Nader Hoveidae, Behzad Rafezy Sahand University of Technology, Faculty of Civil Engineering, Sahand, Tabriz, Iran e-mail: hoveidae@gmail.com; nader.hoveidae@polymtl.ca; rafezyb@sut.ac.ir One of the key requirements for the desirable mechanical behavior of buckling restrained braces (BRBs) under severe lateral loading is to prevent overall buckling until the brace member reaches sufficient plastic deformation and ductility. This paper presents finite ele- ment analysis results of proposed all-steel buckling restrained braces. The proposed BRBs have identical core sections but different Buckling Restraining Mechanisms (BRMs). The objective of the analyses is to conduct a parametric study of BRBs with different amounts of gaps and cores and BRM contact friction coefficients to investigate the global buckling behavior of the brace. The results of the analyses showed that BRMflexural stiffness could significantly affect the global buckling behavior of a BRB. However, the global buckling response occurred to be strongly dependent upon the magnitude of the friction coefficient between the core and the encasing contact surfaces. In addition, the results showed that the global buckling response of BRBswith direct contact of the core andBRM ismore sensitive to the magnitude of contact friction coefficient. Key words: all-steel buckling restrained brace, global buckling, finite element analysis, contact friction coefficient, cyclic loading 1. Introduction Seismic excitations have led to concerns in structural design in earthquake-prone zones. During severe ground motion, large amount of kinetic energy is transmitted into a structure. Seismic codes and studies have been recognized that it is not economical to dissipate the seismic ener- gy within the elastic capacity of materials and, as a consequence, it is preferable to anticipate yielding in some controlled elements. Braces are preliminary devices for energy absorption in braced buildings. However, buckling of the brace in compression results in sudden loss of stif- fness, strength, and energy dissipation capacity. To overcome this deficiency, various types of innovations have beenproposed in steel braces inwhich the buckling has been inhabited through amechanism. Buckling Restrained Braced Frames (BRBFs) have been widely used in recent years. A BRBF differs from a conventionally braced frame because it yields under both tension and compression without significant degradation in compressive capacity. Most buckling restrained brace (BRB) members currently available are built by inserting a steel plate into a steel tube filled with mortar or concrete called conventional BRBs. The steel plate is restrained laterally by the mortar or the steel tube and can yield in compression as well as tension, which results in comparable yield resistance and ductility as well as a stable hysteretic behavior in BRBs. A large body of knowledge exists on conventional BRBs performance in the literature. Black et al. (2002) performed component testing of BRBs and modeled a hysteretic curve to compare the test results, and found that the hysteretic curve of a BRB is stable, symmetrical, and ample. Inoue et al.(2001) introduced buckling restrained braces as hysteretic dampers to enhance the seismic response of building structures. As shown in Fig. 1, a typical BRB member consists 892 N. Hoveidae, B. Rafezy of a steel core, a buckling restraining mechanism (BRM), and a separation gap or unbonding agent allowing independent axial deformation of the inner core relative to the BRM. Numerous researchers have conducted experiments and numerical analyses on BRBs for incorporation into seismic force resisting systems. Qiang (2005) investigated the use of BRBs for practical applications for buildings in Asia. Clark et al. (1999) suggested a design procedure for buildings incorporating BRBs. Sabelli et al. (2003) reported seismic demands on BRBs through a seismic response analysis of BRB frames, and Fahnestock et al. (2007) conducted a numerical analysis and pseudo dynamic experiments of large-scale BRB frames in the U.S. Fig. 1. Components of a BRB The local buckling behavior of BRBs was studied by Takeuchi et al. (2005). The effective buckling load of BRBs considering the stiffness of the end connection was recently studied by Tembata et al. (2004) and Kinoshita et al.(2007). Previous studies have demonstrated the potential ofmanufacturingBRB systemsmade entirely of steel, called all-steel BRBs (Tremblay et al., 2006). In a conventional all-steel BRB, the steel inner core is sandwiched with a buckling restrainingmechanismmade entirely of steel components, thus avoiding the costs of themortar needed in conventional BRBs. This eliminates the fabrication steps associated with pouring and curing the mortar or concrete, significantly reducing manufacturing time and costs. In addition, suchaBRBcanbeeasilydisassembled for inspectionafteranearthquake.Experimental and analytical studies on the deformation performance and dynamic response of BRBs were performed by Kato et al. (2002), Watanabe et al. (2003), and Usami (2006). The restraining member proposed previously was a mortar-filled steel section, which made an extremely rigid member. In such types of BRBs, the brace member and the BRM were integrated, and overall buckling did not occur. However, in all-steel BRBs, which are considered to be a new generation of buckling restrained braces, the brace system ismade completely of steel, and theBRMsystem is lighter in comparisonwith conventional BRBs,which leads to a highpotential for brace overall buckling caused by the low rigidity and stiffness of theBRM.The hysteretic behavior of all-steel BRBswas experimentally investigated byTremblay et al. (2006). An experimental study on the hysteretic response of all-steel BRBs was also conducted by Eryasar and Topkaya (2010). The following characteristics are considered necessary for the safe performance of BRBs: 1) the prevention of overall buckling, 2) the prevention of core local buckling, 3) the prevention of low cycle fatigue of the bracemember, and 4) high strength of the joint parts and connections. In this paper, the first characteristic (i.e., overall buckling behavior) is examined further. Assume a BRB member with initial deflection under compression. When the inner core with initial inherent imperfection deflects under compression, it comes into contact with the BRM. The contact forces increase the out-of-plane deformation of the entire BRB and strength deterioration occurs before the bracemember reaches the target displacement if the rigidity and strength of theBRMare insufficient.According to theAISC2005 guidelines for qualifying cyclic tests of BRBs (AISC 2005), a BRB should undergo axial deformations up to ∆bm, where ∆bm is the brace axial deformation corresponding to the design story drift. The buckling restraining component should have enough strength and rigidity to prevent overall buckling of the brace during axial deformation. Therefore, to obtain the hysteretic characteristic on the compression side similar to that on the tension side and to mitigate pinching, it becomes necessary to avoid Global buckling prevention condition of all-steel ... 893 overall buckling (i.e., flexural buckling). The results of the first studies on overall buckling behavior of BRBs conducted by Watanabe et al. revealed that the ratio of Euler buckling load of the restraining member to the yield strength of the core, Pe/Py, is the factor that is the most determinative for control of brace global buckling(Watanabe et al., 1988). These authors concluded that if the ratio of the Euler buckling load of the BRM to the yield load of the inner core, Pe/Py, is less than one, the brace member will experience overall buckling during cyclic loading of the braced frame. However, a Pe/Py ratio of 1.5 was proposed for design purposes in the studiesmentioned. The criterion Pe/Py ­ 1 has a theoretical basis (Black et al., 2002) and has been verified through experimental testing (Iwata and Murai, 2006). However, the aforementioned studies did not consider the contact properties between the core and the encasing. In other words, the friction coefficient of the core and the encasing contact surface was not considered as an affecting parameter thatmight change the overall buckling prevention condition of a BRB. More experimental studies were conducted by Usami (2006) on all-steel BRBs and a safety factor of λf = Pmax/Py was proposed where Pmax and Py denote the maximum compression force in the brace member and the core yielding capacity, respectively. The safety factor is illustrated as follows γf = 1 Py Pe + Py My (a+d+e) (1.1) where a, d and e are the initial deflection, gap amplitude, and the eccentricity of loading, respectively. Test results showed that if the value of safety factor γf was greater than three, overall buckling of BRBwould not occur. The finite element analysis method was recently used with success to predict the buckling response of the core plates in BRBmembers with tubes filled withmortar (Matsui et al., 2008). Subsequent finite element analysis studies have been conducted by Korzekwa and Tremblay (2009) to investigate the core buckling behavior in all-steel BRBs. The studiesmentioned above also provided a description of the complex interaction that develops between the brace core and BRM. Outward forces induced by the contact forces were found to be resisted in flexure by the BRM components and in the bolts holding together the BRM components located on each side of the core. In addition, the contact forces resulted in longitudinal frictional forces that induced axial compression loads in theBRM.The representative Pe/Py ratio used in these studies was 3.5, and the test results showed that the encasing strength was adequate to prevent global buckling of the brace. Thispapernumerically investigates theoverall buckling inhibition condition of all-steelBRBs regarding the effect of gap size and the friction coefficientmagnitude of the contact between the core and the buckling restraining mechanism. 2. Overall buckling prevention criterion of BRBs Analysis of elastic buckling of a BRB composed of a steel core restrained laterally by a BRM showed that the Euler buckling load of the bracemember under compression could be found by solving an equilibrium equation as follows (Fujimoto et al., 1988) EBIB d2v dx2 +(v+v0)Nmax =0 (2.1) in which EBIB is the flexural stiffness of the BRM, Nmax represents the maximum brace axial load, and v and v0 denote the transverse and the initial deflection of the brace member, respectively, as shown in Fig. 2. 894 N. Hoveidae, B. Rafezy Fig. 2. Force and deformation of a BRB (Qiang, 2005) The initial deflection of the brace is assumed to be expressed by a sinusoidal curve as follows v0 = asin πx L (2.2) where a is the initial deflection of the brace at the center and N is the brace axial load, which is replacedwith P in the following equations. Solving equilibriumEq. (2.1) results in the following v+v0 = a 1− Pmax Pe sin πx L (2.3) The bendingmoment at the center of BRM can be written as follows Mc = Pmaxa 1− Pmax Pe (2.4) where Pmax is the maximum axial force of the brace. Assuming that Pmax is equal to Py (i.e., yield load of the core) and considering that the buckling of the BRB occurs when themaximum stress in the outermost fiber of the BRM reaches the yield stress, the requirement for stiffness and strength of the steel tube (BRM) can be obtained as follows Pe Py ­ 1+ π2EBaD 2σyL 2 B (2.5) in which LB, σy, and D denote the length, the yield stress of the steel tube, and the depth of the restrainingmember section, respectively. This is the first formula that successfully expresses strength and stiffness requirements as paired in the design of BRBs. In this formula, the effect of gap amplitude g has not been considered in the calculation of the moment at the center of the BRM (Qiang 2005). Therefore, in this paper, this parameter is involved in Eq. (2.5). Thus, Eq. (2.5) can bemodified as follows Pe Py ­ 1+ π2EB(a+g)D 2σyL 2 B =β (2.6) where LB is the length of the core and BRM (equal together), and D s the depth of the BRM section. Equation (2.6) indicates that overall buckling of the brace will not occur if the ratio Pe/Py is greater than the parameter β, which is calculated based on the geometric andmaterial characteristics of the brace member. 3. Finite element analysis To provide a numerical understanding of the cyclic behavior and buckling response of all-steel BRBs, a series of finite element analyses were conducted on 24 proposed all-steel BRBs. A three-dimensional representation of the brace specimens was developed to properly capture the expected behavior. The models included the core plates, and the BRM components consist of tubes, guide plates, filler plates, and end stiffeners. Global buckling prevention condition of all-steel ... 895 3.1. Description of the models Finite element analyses have been conducted on 24 proposed all-steel BRBs. Table 1 repre- sents the details and specifications of the models where the second column shows the specimen code in the form S(i)g(j)c(k), in which the indexes i, j, and k represent the model number, the gap amplitude, and the friction coefficient magnitude at the interface, respectively. All of the models consisted of a constant 100mm×10mm core plate with various cross sections for BRMmembers, as shown inTable 1. Therefore, the yield strength of the core was kept constant when the stiffness and strength of the BRMs were altered. The total length of the BRBs, L, was assumed to be 2000mm. The core plate yield load, Pyc, which is illustrated by Py, was calculated by multiplying the yield stress by the cross-sectional area, and the buckling load of theBRM, Pe, was calculated from theEuler buckling load formula. The dimensions of the brace components were selected in a way that the ratio of the Euler buckling load and the yield load of the brace in the specimens, Pe/Py, falls between 1.09 and 2.60. The parameter Ir in Table 1 denotes the moment of inertia of the restraining member. Table 1.Properties of BRB specimens No. Model BRM section Core dimen- Ac Gap Ir Pe Pyc name dimensions [mm] sions [mm] [mm2] [mm] [mm4] [KN] [KN] 1 S1g0c0.1 UNP50+2Fp(45×5) P-100×10 1000 0 81.46E+4 401.99 370 2 S2g0c0.1 UNP65+2Fp(37.5×5) P-100×10 1000 0 111.84E+4 551.91 370 3 S3g0c0.1 B(50×50×3)+2Fp(45×5) P-100×10 1000 0 148.46E+4 732.62 370 4 S4g0c0.1 B(50×50×4)+2Fp(45×5) P-100×10 1000 0 190.00E+4 937.61 370 5 S1g0c0.3 UNP50+2Fp(45×5) P-100×10 1000 0 81.46E+4 401.99 370 6 S2g0c0.3 UNP65+2Fp(37.5×5) P-100×10 1000 0 111.84E+4 551.91 370 7 S3g0c0.3 B(50×50×3)+2Fp(45×5) P-100×10 1000 0 148.46E+4 732.62 370 8 S4g0c0.3 B(50×50×4)+2Fp(45×5) P-100×10 1000 0 190.00E+4 937.61 370 9 S1g0c0.5 UNP50+2Fp(45×5) P-100×10 1000 0 81.46E+4 401.99 370 10 S2g0c0.5 UNP65+2Fp(37.5×5) P-100×10 1000 0 111.84E+4 551.91 370 11 S3g0c0.5 B(50×50×3)+2Fp(45×5) P-100×10 1000 0 148.46E+4 732.62 370 12 S4g0c0.5 B(50×50×4)+2Fp(45×5) P-100×10 1000 0 190.00E+4 937.61 370 13 S1g1c0.1 UNP50+2Fp(45×5) P-100×10 1000 1 85.00E+4 419.46 370 14 S2g1c0.1 UNP65+2Fp(37.5×5) P-100×10 1000 1 116.00E+4 572.44 370 15 S3g1c0.1 B(50×50×3)+2Fp(45×5) P-100×10 1000 1 152.60E+4 753.05 370 16 S4g1c0.1 B(50×50×4)+2Fp(45×5) P-100×10 1000 1 195.23E+4 963.42 370 17 S1g1c0.3 UNP50+2Fp(45×5) P-100×10 1000 1 85.00E+4 419.46 370 18 S2g1c0.3 UNP65+2Fp(37.5×5) P-100×10 1000 1 116.00E+4 572.44 370 19 S3g1c0.3 B(50×50×3)+2Fp(45×5) P-100×10 1000 1 152.60E+4 753.05 370 20 S4g1c0.3 B(50×50×4)+2Fp(45×5) P-100×10 1000 1 195.23E+4 963.42 370 21 S1g1c0.5 UNP50+2Fp(45×5) P-100×10 1000 1 85.00E+4 419.46 370 22 S2g1c0.5 UNP65+2Fp(37.5×5) P-100×10 1000 1 116.00E+4 572.44 370 23 S3g1c0.5 B(50×50×3)+2Fp(45×5) P-100×10 1000 1 152.60E+4 753.05 370 24 S4g1c0.5 B(50×50×4)+2Fp(45×5) P-100×10 1000 1 195.23E+4 963.42 370 B – BOX; Fp – Face plate; P – Plate The core plate and BRM was modeled using 8-node C3D8 brick elements. Large displace- ment static cyclic analysis was performedusing theABAQUS6.9.3 (2005) general-purpose finite element program.The core plate was expected to undergo large plastic deformations and higher mode buckling with pronounced curvature. Therefore, a refinedmeshwas adopted with five ele- ments across theplate and twoover the thickness.A coarsermeshwasused for theBRMbecause 896 N. Hoveidae, B. Rafezy most of this component was expected to remain elastic. Contact properties with hard stiffness in the transverse direction and tangential coulomb frictional behavior were assumed between the core and the BRM. Regarding studies in the field (Chou and Chen, 2010), a coefficient of friction of 0.1 was adopted to provide a greasy interface between the core and the BRM in the models. In addition friction coefficients of 0.3 (Korzekwa andTremblay, 2009) and 0.5 were also adopted to provide a smooth and rough steel-to-steel contact surfaces between the core and the encasing, respectively. The contact model allowed for the separation of the core plate from the BRM element, which enabled the higher mode buckling of the core plate. The core plate and the BRM components were made of steel with a yield stress of Fy = 370MPa. Young’s modulus of 200GPa and Poisson’s ratio of 0.3 were assumed for the core plate and the BRM components. A nonlinear combined isotropic-kinematic hardening rule was employed to reproduce the inelastic material property and, therefore, accurate cyclic behavior. The initial kinematichardeningmodulus C and the rate factor γwereassumedtobe 8KN/mm2 and75, respectively (KorzekwaandTremblay, 2009).For isotropichardening,amaximumchange in yield stress of Q ∞ =110MPaanda rate factor of b=4were adopted.An initial imperfection of 2mm (i.e., L/1000) was considered in both the core plate and the BRM. Two types of interfaces between the coreplate andBRMwere considered in themodels. In thefirst case, direct contact of the coreplatewith theBRMwas implemented, and in the second case, gap amplitudes of 1mm were provided through the core thickness. In addition, a constant gap amplitude of 2mm was provided through the core width (on both sides) in all models. Such a gap was used to accommodate the free expansion of the inner core under axial loads. The axial deformation was blocked at one end of bracing with a pinned connection. Axial displacements were imposed at the other end following the cyclic quasi static protocol suggested by AISC seismic provisions for BRBs (2005) as follows: 2 cycles at ±∆y, 2 cycles at ±0.5∆bm, 2 cycles at ±∆bm, 2 cycles at ±1.5∆bm, and 2 cycles at ±2∆bm, where ∆y is the displacement that corresponds to the yielding of the core, and ∆bm is the axial deformation of the brace corresponding to the design story drift. Based on the previous studies by Tremblay et al. (2006), the peak strain amplitude in full-length core braces typically falls in the range of 0.01 to 0.02 for common structural applications, and the peak deformation in themajority of past test programs have been limited to that range (Watanabe et al., 1988). In this study, ∆bmwas set to 20mm,which corresponds to theaxial strainof 1% in thecore, andthecoreyieldingdisplacement ∆y was calculated as3.7mm based on thematerial characteristics. Therefore, the ultimate axial displacement demand of the brace during cyclic loading will be 2∆bm = 40mm, which corresponds to a core strain of 2%. Therefore, the adopted value for the peak strain demand of the inner core seems reasonable. A typical cross section of the proposed BRB member and its finite element representation are shown in Figs. 3 and 4, respectively. Fig. 3. Typical cross section of proposed BRBs Global buckling prevention condition of all-steel ... 897 Fig. 4. Finite elementmodel of a proposed BRB 4. Results and discussions Hysteretic responses in all of the BRBmodels are well predicted by the finite element model in both elastic and nonlinear ranges. Figuree 5 and 6 illustrate the normalized hysteretic responses of the braces. In the curves, compression is positive. Fig. 5. Hysteretic responses of the proposed BRBs including direct contact of the core and BRM 898 N. Hoveidae, B. Rafezy Fig. 6. Hysteretic responses of the proposed BRBs including a gap between the core and BRM Axial force-displacement curves of the BRB models are captured from a point at the brace end. This point is located in a region that essentially remains elastic because stiffener plates are provided to prevent local buckling in the brace end. Therefore, the captured force-displacement relation may not be a representation of the true stress distribution of the core during cyclic loading, although the curves properly describe the deterioration in strength caused by the global or local buckling of the brace. The axial force-deformation curves shown inFigs. 5 and 6 indicate the sudden deterioration in the strength and overall buckling only in the models S1g0c0.1 and S1g1c0.1 with lower values of Pe/Py among themodels with the friction coefficient of 0.1 at the interface, whereas, in all of the othermodels with friction coefficient of 0.1, the stable hysteretic responsewithout a significant change in the brace load carrying capacity is specified. In addition, in the similar models with the friction coefficients of 0.3 and 0.5, a premature overall buckling is observed which can be deducted from the hysteretic curves in Figs. 5 and 6. All of themodels including direct contact with the friction coefficients of 0.3 and 0.5 experience global buckling during a cyclic loading up to 2∆bm as shown in Figs. 5 and 6. In addition, all of the models with the friction coefficient of 0.3 and 0.5 and containing a gap size of 1mm through the core thickness experience overall buckling except models S4g1c0.3 and S4g1c0.5 with larger strength and stiffness of BRM. The buckled shape of the brace is represented in Fig. 7a. The values of Global buckling prevention condition of all-steel ... 899 Pe/Py have been calculated for all 24 BRB specimens and are given in Table 2. In addition, the representative parameter β is calculated and shown in Table 2. Fig. 7. (a) Overall buckling of model S1g0c0.1; (b) comparison of frictional dissipated energy in model S4g1c0.1, S4g1c0.3, and S4g0c0.1 Table 2.Analytical results for the proposed BRBs No. Model α= Pe Py β Global name [8] buckling 1 S1g0c0.1 1.09 1.11 Yes 2 S2g0c0.1 1.49 1.13 No 3 S3g0c0.1 1.98 1.15 No 4 S4g0c0.1 2.53 1.15 No 5 S1g0c0.3 1.09 1.11 Yes 6 S2g0c0.3 1.49 1.13 Yes 7 S3g0c0.3 1.98 1.15 Yes 8 S4g0c0.3 2.53 1.15 Yes 9 S1g0c0.5 1.09 1.11 Yes 10 S2g0c0.5 1.49 1.13 Yes 11 S3g0c0.5 1.98 1.15 Yes 12 S4g0c0.5 2.53 1.15 Yes 13 S1g1c0.1 1.13 1.24 Yes 14 S2g1c0.1 1.55 1.26 No 15 S3g1c0.1 2.04 1.30 No 16 S4g1c0.1 2.60 1.30 No 17 S1g1c0.3 1.13 1.24 Yes 18 S2g1c0.3 1.55 1.26 Yes 19 S3g1c0.3 2.04 1.30 Yes 20 S4g1c0.3 2.60 1.30 No 21 S1g1c0.5 1.13 1.24 Yes 22 S2g1c0.5 1.55 1.26 Yes 23 S3g1c0.5 2.04 1.30 Yes 24 S4g1c0.5 2.60 1.30 No [8] – Fujimoto et al. (1988) Based on the results of analysis and as shown in Table 2, models with a Pe/Py ratio greater than 1.2 and the friction coefficient of 0.1 donot experience overall buckling during axial loading up to a core strain of 2%. In addition, in these models, the Pe/Py ratio is greater than the parameter β. Therefore, the analysis results confirm the validity of Eq. (2.6). Moreover, the buckling prevention condition (i.e. Pe/Py ­ 1.2) is not dependent on the gap size between the core and the encasing member in the models with the friction coefficient of 0.1. Table 2 shows 900 N. Hoveidae, B. Rafezy that themodelswith higher friction coefficients, such as 0.3 or 0.5, and including adirect contact of the core and theBRMexperience global buckling despite of owning a larger Pe/Py ratio such as 2.53. In addition, the models with the friction coefficients of 0.3 and 0.5 and containing a gap of 1mm at the interface endure overall buckling when the Pe/Py ratio is less than 2.6. The reason is that, in themodels with higher friction coefficient, the slippage of the steel core inside theBRMdoes not occur freely and the applied axial displacement causes the brace (with initial imperfection) to deform laterally instead of free axial deformation. In addition, a part of frictional forces developed at the interface is transmitted into the BRM, which causes the lateral deflection of the brace because of P-∆ effects. Therefore, the overall buckling behavior of theBRBmembers is dependent on the brace interface detail proper- ties especially the magnitude of the friction coefficient. Frictional dissipated energy in models S4g1c0.1, S4g1c0.3, and S4g0c0.1 is compared together in Fig. 7b. As shown in Fig. 7b, the BRB with the direct contact of the core and BRM own larger frictional dissipated energy in comparison to the BRB including the gap. In addition, the frictional dissipated energy in the BRBmodel with the higher friction coefficient is larger. During cyclic loading of aBRBmember containing a gap between the core and the encasing, the brace member causes lateral deflection as the compressive displacement increases and the lateral deflection rises.Contact forces acting on theupper side of theBRMincrease, andbuckling of thebracemember occurswhen themoment at the center of theBRMas a result of the contact forces reaches the yieldmoment of theBRM. Inmodels containing the gap, the lateral deflection rises to deformation of higher order buckling modes while enforcing compression displacement. The contact forces acting on both sides of the restraining member increase under compression and cause global buckling of the brace. The results show that the models with a Pe/Py ratio greater than 1.2 and the friction coefficient of 0.1 do not experience global buckling regardless of the size of the gap. While loading the BRBs including the gap, severe inelastic excursions occur in the core plate under compression, which induces lateral opening of the BRMmember. Previous studies conducted by Korzekwa and Tremblay (2009) also confirm this phenomenon. The results show that the overall buckling behavior of BRB models with direct contact of the core and the BRM is more sensitive to frictional response at the interface. In the other words, in the models without a gap at the interface, the frictional forces developed between the core and the encasing contact surface are considerably larger in comparison to those in the models including the gap. The excessive frictional forces generated at the interface result in the large axial force transmission into the BRM, the development of bendingmoments in the BRM, and the overall buckling of the entire brace, consequently. Based on the results, the overall buckling behavior of BRBs depends on the interface detail properties. The known parameter Pe/Py and the magnitude of friction coefficient at the core and the BRM interface are the most effective parameters that influence the overall buckling response of a BRB. Employing an unbonding material such as butyl rubber at the interface provides a surfacewith a friction coefficient near 0.1. In this case, the overall buckling prevention condition of the brace can be similar to the criterion suggested by previous researchers (i.e., Pe/Py ­ 1)(Watanabe et al., 1988). However, further numerical studies and experimental tests are necessary to examine and suggest guidelines on the overall buckling prevention condition and the design of all-steel BRBs with different interface details and various amounts of friction coefficients at the interface, consequently. 5. Conclusions One of the key requirements of buckling restrained braces is the performance of avoiding ove- rall buckling until the brace member reaches target displacement and sufficient ductility. This required performance becomes important as the BRB is lightened, and the strength and rigi- Global buckling prevention condition of all-steel ... 901 dity of the restraining member are reduced. A new generation of BRBs, called all-steel BRBs, is a class of BRBs with lighter buckling restraining components than conventional BRBs. In this family of BRBs, a light steel component is used as a restraining member instead of the mortar-filled tubes used in conventional BRBs, whichmay result in overall buckling of the brace caused by inadequate rigidity and strength of the restraining components. In this paper, the overall buckling prevention condition of all-steel BRBs considering the core and the encasing interface detail is numerically examined through the finite element analysis method. Among 24 proposed all-steel BRB specimens, the models with friction coefficient of 0.1 and a Pe/Py ratio less than 1.2 experienced global buckling during cyclic loading of the brace up to a core strain of 2%, which closely meets the overall buckling avoidance condition of BRBs suggested by previous researchers. However, larger Pe/Py ratios are required to prevent overall buckling of the brace as the friction coefficient between core and BRM is increased. Themain out-com of the study can be summarized as follows: • Results of analysis show that for the BRM component a larger strength and stiffness is required to inhibit global buckling of the bracewhen a higher friction coefficient or a rough surface is specified at the core and BRM interface. The known global buckling prevention condition of BRBs, Pe/Py > 1.2, can be applied only for braces with smooth surfaces between the core and BRM and smaller friction coefficients, such as 0.1 and lower. • The overall buckling response of the BRB is so sensitive to the magnitude of friction coefficient at the interface in the case with direct contact between the core and the BRM. In other words, the overall buckling of BRB with direct contact between at the interface corresponds to larger values of Pe/Py in comparison to the models including a gap since the frictional forces developed at the interface in BRBswith direct contact of the core and BRMare higher in comparison to the BRBs including the gap. This leads to transmission of higher axial forces into theBRMandbendingof the entire brace because of P-∆ effects. • It is recommended to use an unbondingmaterial at the core andBRM interface to reduce frictional forces and avoid premature overall buckling of BRBs. In addition, employing a gap between the core and BRM not only provides enough space to accommodate free lateral expansion of the core but also decreases the frictional forces and the chance of global buckling of the brace for a constant stiffness and strength of the BRM as a result. In addition, the cost of using such a material is low in comparison to the overall cost of fabrication of a BRB. However, although using a rough material between the core and BRMincreases themagnitudeof friction forces at the interfacewhichmay causepremature buckling of the BRB member; it noticeably increases the energy dissipation capacity of the brace by friction. • Theglobal bucklingconditionof aBRBisdependenton the interface characteristics suchas the magnitude of the friction coefficient at the interface and the gap size. Further exten- sive experimental and numerical studies are required to survey and suggest the overall buckling prevention condition of all-steel BRBs for design purposes considering the inter- face properties, especially the contact characteristics between the core and the encasing member. References 1. ABAQUS, Standard user’smanual version 6.3, 2005,Pawtucket,RI: Hibbitt, Karlsson&Sorensen, Inc. 2. AISC (American Institute of Steel Construction), 2005, Seismic provisions for structural steel bu- ildings, Chicago, IL 902 N. Hoveidae, B. Rafezy 3. BlackC.J.,Makris N., Aiken I.D., 2002,ComponentTesting, StabilityAnalysis, andCharac- terization of BucklingRestrainedBracedFrames, ReportNo. PEER2002/08,California, Berkeley, CA 4. ChouC., Chen S., 2010, Sub-assemblage tests andfinite element analyses of sandwichedbuckling restrained braces,Engineering Structures, 32, 2108-2121 5. Clark P., Aiken I., Kasai K., Ko E., Kimura I., 1999,Design procedures for buildings incor- poratinghystereticdampingdevices,Proceedings of 69thAnnualConvention, SEAOC, Sacramento, USA 6. Eryasar M., Topkaya C., 2010, An experimental study on steel-encased buckling restrained brace hysteretic damper,Earthquake Engineering and Structural Dynamics, 39, 561-581 7. FahnestockL.A., SauseR., Ricles J.M., 2007, Seismic response andperformance of buckling- restrained braced frames, Journal of Structural Engineering, 133, 9, 1195-1204 8. FujimotoM.,WadaA., Saeki E.,WatanabeA., HitomiY., 1988,A study on the unbonded brace encased in buckling restraining concrete and steel tube, Journal of Structural and Construc- tion Engineering, 34B, 249-258 9. Inoue K., Sawaizumi S., Higashibata Y., 2001, Stiffening requirements for unbonded braces encased in concrete panels,ASCE Journal of Structural Engineering, 127, 6, 712-719 10. Iwata M., Murai M., 2006, Buckling-restrained brace using steel mortar planks performance evaluations as hysteretic damper,Earthquake Engineering and Structural Dynamics,35, 1807-1826 11. Kato M., Usami T., Kasai A., 2002, A numerical study on cyclic elasto-plastic behavior of buckling-restraining brace members, Journal of Structural and Construction Engineering, 48A, 641-648 12. KinoshitaT.,KoetakaY., InoueK., IitaniK., 2007,Criteria of buckling-restrainedbraces to prevent out-of-plane buckling, Journal of Structural and Construction Engineering, 621, 141-148 13. Korzekwa A., Tremblay R., 2009, Behavior of Steel Structures in Seismic Areas, Taylor & Francis Group, London, Chap. 94 14. Matsui R., Takeuchi T., Hajjar J.F., NishimotoK., Aiken, I., 2008, Local buckling restra- int condition for core plates in buckling-restrainedbraces,Proceeding of 14thWorld Conference on Earthquake Engineering, Beijing, China 15. Qiang X., 2005, State of the art of buckling-restrained braces in Asia, Journal of Constructional Steel Research, 61, 727-748 16. Sabelli R., Mahin S., Chang C., 2003, Seismic demands on steel braced frame buildings with buckling-restrained braces,Engineering Structures, 5, 655-666 17. Takeuchi T., Suzuki K., Marukawa T., Kimura Y., Ogawa T., Sugiyama T., 2005, Per- formance of compressive tubemembers with buckling restrained composed ofmortar in-filled steel tube, Journal of Structural and Construction Engineering, 590, 71-78 18. Tembata H., Koetaka Y., Inoue K., 2004, Out-of-plane buckling load of buckling restrained braces including brace joints, Journal of Structural and Construction Engineering, 581, 127-134 19. Tremblay R., Bolduc P., Neville R, DeVall R., 2006, Seismic testing and performance of buckling restrained bracing systems,Canadian Journal of Civil Engineering, 33, 1, 183-198 20. Usami T., 2006, Guidelines for Seismic and Damage Control Design of Steel Bridges, Japanese Society of Steel Construction, Gihodo-Shuppan, Tokyo, Japan 21. Watanabe A., Hitomi Y., Yaeki E., Wada A., Fujimoto M., 1988, Properties of braces encased in buckling-restraining concrete and steel tube, Proceedings of 9th World Conference on Earthquake Engineering, 719-724 22. WatanabeN.,KatoM.,UsamiT.,KasaiA., 2003,Experimental study on cyclic elasto-plastic behavior of buckling-restraining braces, Journal of Earthquake Engineering, 133 Manuscript received December 5, 2012; accepted for print February 15, 2013