HYSTERETIC BEEAVIOR OF STEEL BRACES SUBJECTED 70 HORIZONTAL LOAD ‘DUE TO EARTHQUAKE by 1 IT ys = Minoru WAKABAYASHT", Tekeshi NAKAMURA™™, Michio SHTBATA’ Nozomu YosxrDA"” ena Hiromi Masupa” SYNOPSIS In this peper, Giscussed are the elastic-plastic hysteretic behavior of steel braces which play an important role as the earthquake resistant elenents of steel structures. Experimentel end theoretical investigations on the braces by the authors are briefly introduced in the first half. In the latter half, ideslized post-buckling curve and hysteresis loops of the brace menbers are formulated and proposed for the design use, based on the parametric analysis of the experimental and theoretical results. INPRODUCTION The earthquake resistant function of steel structures with a vertical, bracing system, much depends on the hysteretic behavior of braces. It is essential to obtain the appropriately idealized restoring force cheracter- istics and hysteretic rules of bracing meubers vhich are usefully applicable to the structural design. ‘The elastic-plastic behaviors of simply supported steel braces under monotonic and alternately repeated loading were investi- gated experinentally by the euthors in Refs. 1), 2), 3) and theoretically in Refs. 5), 6), 7), 8). The behaviors of braces under more realistic condition, were reported in'Refs. 4), 6), 9). A parametric data enelysis for the re~ sults of ebove-mentioned researches was performed to formulate idealized hysteresis loops for a single brace. The loops for peired braces vhose behav- fors do uct interact each other can be obtained by the superposition method. ‘EXPERIVENPAL STUDIES A series of experiments wes conducted to investigate the fundenental. properties of the elastic-plastic behavior of breces. In the first series, tested are sinply supported bers with square cross séction(15x15 mm). Figure 1 shows an exemple of load-displacenent relationships reported at ‘the 5th WEE, in Rome. In the figures, n and 6 are the non-dimensional expressions of the axial force Nand the axial displacement A, respectively, divided ty their yield values; Ny and Ay. ng is the slenderness parancter defined es the ratio of Fuler losd Ng to Ny. Basic characteristics of a brace vere obteined fron this experimentel study. In the second series of experiments, vide flange braces(H-50x50x6x6) in @ single or double bracing system vere tested under more realistic condition. Main interestings were framed effects and the effects of local instability of section elements on the hysteretic behavior. Test set-up end specinen are illustrated in Fig: 2(a) and (b), respectively. Figs. 2(c) and (a) show examples of loed~dis~ Blecenent curves. Mein informations from the test are as follows: (e) Hys~ ‘teresis loops unter the preseribed displacement amplitude stabilize more rapidly in the case of strong axis buckling then week axis buckling. How- ever, after a few cycles of loading leterel deflection becomes predominant about a weak axis. (b) The behavior of © single brace specimen designed to ‘Duckle about its veak axis in plane of the frame is almost the same es thet designed to bustle out of plane of the frame, although the boundary condi Professor, Disaster Prevention Research institute, Kyoto University. Tr dAesistene) Disaster Prevention Research Tastitute, Rot University, II Lecturer, Department of Architecture, Osake Institute of Techmology. IV Graduate Student, Kyoto University, 'V Engineer, Naito Design Office. 3188tions of them are not the same strictly. (c) The effects of seconim bend~ ing on the behavior of braces due to the existence of another frane ele nents are not important. (4) Flange local buckling at both ends and mid~ span initiated by lateral deflection induces the concentration of plastic deformation and causes fatigue cracks at these portions. THEORETICAL STUDIES Two types of theoreticel analyses on the behavior of the bar subjected to repeated axial force were reported at the Sth WCEE, in Rome, Theory (2) treated a ber as one dimensional continu based on the elastic-perfectly plastic theory. Theory (II) vas developed on the basis of modified Shanley Hodel. The foruer vas extended to investigate the behavior of the bar with various types of yield conditions and estimate the effects of elastic ent restraints ageinst rotetion, The latter was extended to examine the effects of constraints of end displacenent of © brace by the frane elenents. From the anelysis, the behavior of the brace rigidly connected to frane elenents with ordinary dimensions could be epproxinsted to thet of clanped brace or ‘thet of equivelent simply supported brace with « half length, A numerical analysis which takes into account of the effects of partial yielding of « section and extension of yield zone in longitudinal direction using nore realistic stress-strain relationship under cyclic loading was nevly done to trace sccurately the experimental load-displacenent relationship end to supplenent the lack of the experinentel date under rendon losding process in the formulation of the Loops. YATHEMATICAL EXPRESSION OF HYSTERESIS LOOPS The formulstion procedure of the mathenatical expression of hysteretic oops is presented here. The formulated loops should be applied to the brace in the range of slenderness ratio 30150 with rectangular or wide~ flange cross section designed to buckle avout its week exis. General Description of Hysteresis Loops The hysteresis loops of a single Brace can be characterized to be composed of four steges shown in Fig. 3(a). Stage A : The ber is almost straight ond load capacity is nearly equal to the yield load in pure tension. Stage B : The portion near the mid-span of the bar is yielding under combined stress in tension and bending. Stage C The portion near the mid-spen of the ber is yielding in compression end flex- ure. Stage D: Elastic unlosding range. These properties are idealized ty a pair of mechanism Lines end linear elastic renge in the model loop, as shown in Fig. 3(b). It is assumed thet any compression or tension mechanism Lines keep the prescrived shapes in whole process of repested loading, and can be obtained by the translation of a standard compression or tension me chanism line in é-axis direction, respectively. Tae slope of the elestic recovery Line is determined fromthe assumption that 6,/6¢ = 5,'/B¢' = const. (see Fig. 3(€)). When the loading direction is reversed at point £ on the tension yield line (Stege A), the point representing the resyonse of « ber moves tovard point 0 on the elastic recovery line with the slope identical to the initial one (Stage D). From point 0, the response poirt moves on the compression mechanism line (Stege C). Amouzt of the horizontal movenent of the responce point on the compression mechanisn line prescribes the trans~ lation of point P. When the losding direction is reversed at point 1 on the compression mechanism line the response point moves on the elestic recovery line tovard point 2 from vhere it moves on the nev tension mechenisn line, specified by poin! P! (Stege B), while point 0 trenslates to its nev posi- tion 0". If the response point ‘passes through point P', it moves on the tension yield line. On the other hand, if the loading direction is reversed at point 3 on the tension mechanism line, the response point moves on the 3189new compression mechanism Line specified by point 0 after passing through the elastic recovery line. Determingtion of Compression Mecheniom Line Tue post~buckling behavior of @ simply supported ber with ideal I-section made of elastic-perfectly Plastic material was derived as the following equation in Ref. 7). 6 =n (1 = n)(ng = n)/(3n) + ng(2 ~ 2)*/(6n*) nis expressed as the following form. n= 1/¥éi/ag +1 + 2/ng - 8/ng/60/np + 1 + G/ng - Og Based on this equetion the compression mechanism line for rectangular cross section is approximated to n = 1/(a+d + >)", Fig. (a) shows 1/n* vs. & relations obtained fron« sents, These relationships are approximated by straight lines and their slopes and intersections vith the ordinate correspond to @ and b, respectively. @ and b are determined trom Fig. 4(b). n= if(aé +b)" , a(ng) = (10/ng-1)/3 2.0 , b(ng) = K/ng + 0.6 21 (1) ‘The formulated curves are compared with experimental ones, in Fig. &(c). Determination of Tension Mechanism Line The tension mechanism line is deliberately put in the Same form as the compression mechanism line, n=i/(e-6+a)¥, Where d = 1 from the definition that the mechanism lint should pass through the normelized point (0,1) in Fig. 5(c). As shown in Fig. 5(a), (1/n%®-1) vs. 6 relation obtained fron experimental tension mechanism lines are epproxineted to straight lines which pass through the origin, Coefficients e and r are determined fron Fig. 5(b), as follows. neaiffes+i)F 5 r= 3/2, ¢=1/(3-2 ng +1-4) (2) In Fig. 5(c), Eq. (2) is compared with experimentel results. Translation Rule of Mechanism Lint In Fig. 6(a), plotted ere the trans- Ietion x of the tension mechanism line under constant displacement amplitude 8, versus the horizontal distance s between the latest characteristic point P ahd the load reversal point Q relations (see Fig. 3(c)). From the figure, it 4s understood thet x-s relations are linear and the slope of the curves is independent of the displecenent amplitude. From ebove-mentioned con- sideration, x is approximated to x = % ~ n-s, where n = 0.115/ng + 0.36 from Fig. 6(b). '% corresponfing to the transistion in the case of s = 0 is deter~ mined as ihe function of the plastic displacenent amplitude 5, (see Fig.3(4)). xe anlt-3g 42) - ne, t= (3 2/ap)/0 (3) ‘The compression mechanism lines translate in such menner thet the condition 8) /69' = 64/8 is satisfied, in Fig. 3(b) Blestic Recove: © Figure T(a) shows 6, vs. 6, relations (see Fig. 3 Gah) obtained fon the experinents. The ratio'E = 6,6, ts approxinated to ‘ve independent of the displacement amplitude, and is derived from Fig. 7(b), a4 follows, & = 0.3% + 0.2h22 (4) ‘The formated loops illustrated in Figs. 8(e) end (>) approximate well, ‘the experimental behaviors shown in Figs. 8(c) and (a), respectively. MATHEMATICAL EXPRESSION OF POST-BUCKLING CURVE UNDER MONOTONIC LOADING a> this section, derived is the mathenstical expression for the post- buckling vehavior of a brace under compressive axial force. The n vs. 6 re- Jetionship is simplified to the same form as the one in the previous sec- tion, accompanied by three modification functions g), 82 and 63- 3190n= gy Bs 65/(a'-d + bt) Sa a! and b' for a brace vith ideal T-section are given by the correletion for theoreticsl results in Ref. 7) a'(ng) = (1/ng + 6)/(2ng) 5 (np) © 1 + 2/ag The function g, which estimates the accurate buckling strength te given that 81 = Mop(mg): {a's non(np) + vt} (6) using arbitrarily selected colum-curve noy(ng). Simple expression of the Junction function gp is given by 82 = (1/gg - 1/e))-nep(mg)/S + 1/ey (7) The modification function g3 to take into account of the effects of the shepe of cross-section on the post-buckling behavior is given by the follow ing equetion as the approximation curve for the theoretical results in Refs. 6) and 9) (Fig. 10). G, = 2a-(1- al-np/é + 0.60 + 0.4 $2.5 (8) Figure 11 shows the formulated post-buckling curves. In Fig. 12, the formated curves are compared with the experimental results using the col- umn curve specified by the AIJ design standard for steel structures. CONCLUDING REMARKS (2) Based on a grest store of information from experimental and theoretical works on the elastic-plastic behaviors of steel braces under cyclic loading, formulated vere the accurate mathematical expression of the post-buckling curves and the hysteresis loops which can be directly applied to the dynamic anelysis or structural design of bracing syst (2) The formulated loops in which the load capacity vas expressed as the func~ ‘tion of displacenent vere characterized by four basic equations; two mecha~ nism lines and two hysteretic rules, when the slenderaess ng vas settled. (3) In the case of monotonic loading in compression, more accurate expression of post-buckling behavior vas formulated for braces with various types of cross section.” In this formula, the column curve(ne;-ng relation) which is appropriately selected as the needs of the case denands, can be adopted. (4) The hysteresis loops and the post-buckling curves proposed for the single bracing system, can be easily applied to the design use of the double- or multi-brecing system using the superposition method. REFERENCES 2% brea, 2 t,o, Khe a aso, "ho Rees ce te Raa of «Snel er ude Aepeted Ural Loustg’ amualny Disescer trevetion Reser Trt tste, Rte Cavey, or ahs pe 7 = Seon an sages): = 2) Ws akbarahs, Boa, 3. Haber, 8, vorio al Yshidn, “apeinestal tuts onthe Bebrioe of Seed tarts fepeees sal tnsing Fat 1% fala, iseier Poverdion Resear Sanita ato al- selieys fos lah, gee 34d 15, 918 Un Sauna 2) He mibte, T Memsrk, , Yoma, S, Moron, T-Honaa aM Noeaoht, “East t-Platie Deharior of Seed batt tor Repectes Mun eading's ber of the oth Wty py us =Blbs honey 19T- 1) prof. uabeyahi's Ub, "perieenad uit onthe Barocas ltartFroerts of the Atul Braces", pore to tbe blicig Ravens fntiane, Maty of Sasration, Yoke Ue Sr HOH AME tin Seana) 5), Nonaka, "he musetePlaete Anzete of « ar water Noel Atal Leeog", Tet. 2, solide Strstern, Tar drop 9 7 or sy wth eae tt 60% atbauan, 7, Yorba aol K.Younidn, “he Asie of « br lth el ettratate ander Rept Ae Lode Tufe avttcts, somos Mating of st artes Of RS, pr SOT Be ite ie Sepa) 7 We Webern, Bnet K Seba, "Stes onthe Fostuking bebevion cE ree Pare 2, Aatrste, Moon Ween of His Renn se Lay phe DoT = tye ts Sepa) © me eben, enbece at Manda, hn ase i-Paede epee of hr oubgectd teat Eplaceers", Brant Etaster Prevention starch istitte, Wes Onvesti. B, Sey poe ROT Ese to a Sepa 9) M taasaeans asd Ms Since, "sties on toe Tove-Dsthng tberior of Hacen, Part", Aeetran Seeig of Kata ot AS, ppe ak = iy pT Cn Spenae 3191(a) Bod Support. (bo) np = 1.13 (A = 85). Hig. 1 Test of Simply Supported Bar. - . EF "DR Test Set Up. (v) Specimen. (a) Double Bracing system. Fig. 2 Test of Wide Flange Brace. (a) 1/2? vs. 8 (b) 2,0 vs. ng. (c) Formulated Curves. Fig. | Formuletign of Compression Nechanian Line. ensiet wr TO (a) (a/n% ~ 1) vs. 6. (b) I/c vs. ng. (ce) Formulated Curve: Fig. 5 Formulation of Tension Mechanism Line. 3192(c) Beperinent. (a) Experinent. Fig. 8 Comperison of Formulated Loops with Experiments. : A ipa (a) ng = 0.53 (b) ng = 1.96 (c) mpl) ve. 5y, Fg. 9 Formulated Loops for Double Bracing Systen. + A Fig. 11 Formulated Post~Buckling curve. + (o) & v8. Sap (e)ng= 20 i ————— on 5 Fig. 7 Elastic Fig, 10 Effects of Fig. 12 Comparison of Formulated ied cross Sectional Shape. Carve with Experiment. 3193DISCUSSION S.C. Goel (U+S-As Would you elaborate on how your mathematical model. accounts for the so-called "Column growth" phenomenon, i-e+, the deterioration of tensile and compressive strength of a single brace under constant amplitude cyclic deformation ? Author's Closure With regard to the question of Mr- Goel, we wish to state that in the mathematical model, the deterioration of load carrying capacity in tension or compression range is prescribed by the translation x along the deformation axis (abscissa) of the tension or the compression mechanism line, respectively- ‘The compression mechanisn line is a characteristic curve for the post-buckling behavior of a brace and the ten- sion mechanian line is for the post-yielding behavior in the restretching range after buckling. ‘the translation x of the tension mechanism line is given by Eq. (3), using and s, where denotes the deformation amplitude on the latest compression mechanism line and s is a distance between the latest characteristic point P and the load reversal point @ on the tension mechanism line. The coefficients and in Eq. (3) are determined by the parametric data analysis for experimental results (Fig-6)- On the other hand, the translation of the compression mechanism line is prescribed by the relation referring Fig. 3(b)- 3194