1 Enhanced Smooth Hysteretic Model with Degrading Properties 2 Tathagata Ray1 S.M. ASCE and Andrei M. Reinhorn2 F. ASCE 3 4 Abstract 5 The elastic and inelastic nonlinear behavior of structural members use constitutive relations in 6 which restoring forces and deformations are not proportional, and often follow different paths in 7 loading and unloading using hysteretic functions. There are numerous available models that can 8 trace the stiffness and strength changes through yielding, softening and hardening; however, 9 models that can address more complex behavior such as degradations, large deformation, bond 10 slip and joint gap, do so by complex polygonal rules or smooth continuous functions describing 11 momentary (tangent) behavior. There is a need for a unified model based on a combination of 12 mechanical springs that can trace the instantaneous combined stiffness used in system analyses. 13 In this study, a one-dimensional smooth hysteretic model using series and parallel springs, 14 designed for nonlinear structural analysis, is enhanced to incorporate: a) time independent 15 properties, b) nonlinear elastic and post-elastic softening and hardening, c) sudden or continuous 16 variation of strength, d) degradation of elastic and inelastic stiffness, e) a modified bond-slip 17 model, and f) an alternative joint-gap model with a variable gap closing length. Using spring 18 analogues or reciprocal structures, the instantaneous force-displacement incremental relations are 1. Ph. D. Candidate, 224 Ketter Hall, Dept. of Civil, Structural and Environmental Engineering, University at Buffalo, SUNY, Buffalo, NY 14260, USA. email: [email protected] 2. Clifford C. Furnas Eminent Professor, 135 Ketter Hall, Dept. of Civil, Structural and Environmental Engineering, University at Buffalo, SUNY, Buffalo, NY 14260, USA. email: [email protected] 2 19 formulated, which can be further used to determine the instantaneous tangent-stiffness matrices 20 of the elements used in the analyses of complex structures. 21 22 Keywords:-Hysteresis, Nonlinear, Hardening, Degradation, Strength, Stiffness, Pinching, 23 Asymmetric, Geometric 24 Acknowledgements 25 The funding for the research was provided by the National Science Foundation under 26 Grants CMMI-NEESR: #0721399 and #0830391. The authors are also grateful to Ki Pung Ryu 27 (PhD Candidate, University at Buffalo) for his valuable feedback. 28 3 29 Introduction 30 Structural elements in general show nonlinear force-displacement (or moment-curvature) 31 relationships, accompanied with either hardening or softening at larger displacements (or 32 curvatures). This nonlinear behavior can be coupled with degradations in terms of strength and 33 stiffness, bond-slip-pinching, asymmetric yielding and gap closing. Nonlinear behavior arises 34 from (i) material nonlinearities or (ii) geometric changes. Material nonlinearity causes hysteresis 35 in the force deformation relationship when sinusoidal or random loading is applied to a structural 36 element. Mathematical modeling of this hysteresis behavior is usually done either through a 37 Polygonal Hysteretic Model (PHM) or a Smooth Hysteretic Model (SHM), as described by 38 Sivaselvan and Reinhorn (1999). 39 predefined lines that fit observed behavior. The smooth models usually track the stiffness 40 changes through a combination of component springs (reciprocal structures) with 41 phenomenological meaning. The polygonal models track the stiffness changes along 42 Clough and Johnston (1966) presented a PHM where the load-displacement path after a 43 load-reversal targets the yield point in a negative direction, if the element hasn’t yielded on the 44 negative side yet, or aims at the previous maximum displacement point attained in the negative 45 direction. A similar rule is applied after load-reversal from a negative displacement increment. 46 Takeda et al. (1970) proposed a PHM considering cracking and yielding with a number of load- 47 displacement “conditions” and branch stiffness “rules.” Park et al. (1987) and Sivaselvan and 48 Reinhorn (1999) extended the Clough and Takeda models to create a generalized PHM with both 49 vertex oriented and yield oriented behavior, and coupled with degradation, pinching, and 50 asymmetric yielding. Iwan (1966) proposed a PHM for an elastic-perfectly plastic element where 51 the elastic branch is piecewise linear. In his model, n number of spring sets (linear spring in 4 52 series with slip elements having finite yield force) are considered in parallel. Iwan (1966) and 53 Thyagarajan (1989) extended this model by using an infinite number of springs, and thus 54 obtaining a curvilinear or smooth hysteretic behavior. Ibarra et al. (2005) proposed a PHM that is 55 also capable of simulating strength and stiffness deteriorations, pinching, and residual strength. 56 Smooth hysteretic models seem to represent the continuous changes in the materials 57 better. A class of smooth hysteretic models (SHM), originally proposed by Bouc (1967) and 58 later modified by Wen (1976), served as the basis for many developments of models (for 59 example, Park et al., 1987; Kunnath and Reinhorn, 1994) that can address the strain rate 60 dependent properties of materials, based on plasticity theory as described by Sivaselvan and 61 Reinhorn (2001). Moreover, Sivaselvan and Reinhorn (2001) developed a more versatile model 62 that can (i) address degradation of strength and stiffness and (ii) track bond slip and gap opening 63 and closing in cross sections of structural members, usually represented by “pinched” hysteretic 64 models by changing the properties of component springs. Rodgers et al. (2012) recently 65 proposed an SHM, formulated through rational mechanics, that can simulate asymmetric 66 yielding, pinching and flag-shaped hysteresis, which can be used to simulate various rate 67 dependent dampers. In the present paper, the formulations of Park et al. (1987), Kunnath et al. 68 (1997) and Sivaselvan and Reinhorn (2001) is extended, using analog springs, in parallel and in 69 series to: (i) simulate rate independent hysteretic behavior; (ii) include nonlinear elastic and 70 post-elastic behavior due to kinematic hardening or softening at large deformations due to 71 structural confinement, or from coupling between lateral and axial behavior at very large 72 deformations, etc.; (iii) incorporate stiffness degradation of a linear elastic spring component, 73 similar to Ibarra et al. (2005), who included such a feature in a PHM framework; (iv) modify the 74 Gaussian “pinching” model to overcome its various drawbacks and alternatively propose a model 5 75 based on a transcendental function that can address bond slip and variable gap closing; and (v) 76 incorporate changes in hysteretic strength in the instantaneous (tangent) stiffness used in the 77 global analysis of nonlinear structures. The properties of the component springs can be derived 78 from mechanical principles, or validated or calibrated with experimental data. All of the above 79 extensions as well as the basic SHM are re-formulated as time independent functions that can be 80 used in both nonlinear static and dynamic analyses. 81 Formulation of the Smooth Hysteretic Model 82 The development of the enhanced model is based on a re-formulation of the original model of 83 Sivaselvan and Reinhorn (2001). This formulation is partially repeated here, with minor 84 modifications and new notations, in order to define the components used in the new 85 developments, which are presented later in the paper. 86 The original model was developed along Bouc-Wen's formulation (Bouc 1967, Wen 87 1976), which defines the restoring force (F) in a nonlinear spring element for a displacement (u) 88 as: 89 F akou 1 a ko Z 90 Here ko is the initial elastic stiffness and a is the post-elastic stiffness ratio. The hysteretic 91 variable Z is defined by the temporal differential equation in the second part of Equation (1), 92 where the derivatives represented by the dot are with respect to time and A, β, γ are three 93 constant parameters. Note that given the constants a, A, β and γ, the two components in Equation 94 (1) may be dimensionally inconsistent. Constantinou and Adnane (1987) showed that in order 95 to satisfy plasticity theory, the following relations must hold true: A=1 and β+γ=1. where Z Au uZ Z uZ 2 (1) 6 96 The model of Sivaselvan and Reinhorn (2001) is reformulated here by re-defining the 97 hysteretic variable Z as a time independent non-dimensional variable: Z = Fh/Fhy. Z helps to 98 describe the load reversal in the plastic range. Here Fh and Fhy are the force functions of the 99 hysteretic spring, which are dependent on displacement and yield strength, respectively. The 100 generalized force-displacement relationship, F-u, for a system with yield strength Fy and yield 101 displacement uy is: 102 F Fy a (u / u y ) (1 a) Fh / Fhy 103 Equation (2) shows that the total force F at any deformation can be expressed as the sum of the 104 two component springs acting in parallel as shown in Figure 2: (i) the basic elastic spring with 105 linear stiffness of the post-elastic component akou, and (ii) the hysteretic elastic ideal plastic 106 spring Fy(1-a)Z. Note that Z is the non-dimensional hysteretic variable, which is defined using 107 the configuration shown in Figure 3 (a) and from Equation (2) as: 108 Z Fh Fhy F akou [1 a Fy ] (2) (3) 109 In the elastic range, Z varies linearly with u, while in the post-elastic range, Z remains 110 constant. This dependency of Z vs. u is also shown in Figure 3(b). The variation of Z with the 111 deformation u can be expressed as: 112 dZ / du 1/ u y 1 Z 113 N is the exponent which controls the “smooth” transition between the elastic and post-elastic 114 branch, hence the name “Smooth Hysteretic Model.” For smaller values (2<N≤10), a gradual 115 smooth transition is produced, while larger values of N (>10) lead to a sharp transition, such as in 116 the bilinear model. The signum function sign(x) is defined by sign(x)=x/|x| for x≠0 and sign(x)=1 117 for x=0. In the elastic range, it is defined as |Z|<1, |Z|N0 for N>2, as seen in Equation (4). N 0.5 sign(Z .du) 0.5 (4) 7 118 Hence in the elastic region, dZ/du=1/uy, and is constant. In the post-elastic range, |Z|=1 and 119 sign(Z.du)=1, hence dZ/du=0. The signum product Z.du controls the load reversal as shown in 120 Figure 3(b). The above model conforms to the kinematic hardening rule described by Chen and 121 Han (2007). 122 In order to obtain the instantaneous force F during analysis for a given displacement 123 history, the first-order differential Equation (4) has to be solved along with Equation (3). 124 Equation (4) can be solved in closed form for N=1. For larger values of N (N>2), a direct 125 solution of Equation (4) is not possible and numerical integration schemes using iterations or 126 semi-implicit integration schemes can be used. The semi-implicit Rosenbrock Integration 127 (Nagarajaiah et al. 1991a, b) is shown in detail in Figure 4. The Rosenbrock integration in this 128 paper uses Rosenbrock’s coefficients a1 and b1, 0.7886751 and -1.1547005, respectively. The 129 coefficients were originally derived by Nagarajaiah et al. (1991a, b) to maintain a fourth order 130 truncation error in the Rosenbrock integration method (Rosenbrock 1963). The factors 0.75 and 131 0.25 in Equation (i) (see Figure 4) are also selected to achieve the same order of approximation. 132 With N being small, i.e., 2 or 3, when the system is loading, unloading and subsequently 133 reloading at low force levels, the model may fail to satisfy the Drucker postulate of loop-closing, 134 as discussed by Thyagarajan (1989) and Sivaselvan and Reinhorn (2001). For this reason, it is 135 prudent to make N larger, i.e., greater than 5 or preferably 10. However, higher values of N may 136 create numerical instability in the solution and the force deformation path may overshoot the 137 yield surface. Therefore, if the absolute value of Z1 or Zi becomes (numerically) larger than 1, it 138 is forced to 1 according to the signum function sign(Z) in order to obtain acceptable results. 139 Casciati (1987) 140 additional term for N=1. proposed another solution to satisfy the Drucker postulate by adding an 8 141 For the nonlinear analyses of structures, the continuously changing instantaneous tangent 142 stiffness ktangent must be evaluated, which can be obtained by differentiating Equation (2): 143 k tan gent (dF du ) kh 1 Fh Fhy 144 Note that Equation (2) includes the value of Z= Fh/Fhy, which is also variable. The tangent 145 stiffness, ktangent, can be evaluated at the beginning of any new computation interval (step) using 146 the value of Z either a) from the previous step, or b) by solving the first order differential 147 Equation (4) for the updated value of Z using a standard semi-implicit solution (as shown in 148 Figure 4), therefore providing a better approximation for ktangent. 149 Stiffness Degradation of the Hysteretic Spring: Based on experiments and observations, the 150 stiffness degradation can be modeled by using the assumption that the unloading branch of the 151 hysteretic spring from a positive yield direction at every unloading targets a pre-defined pivot 152 point (Park et al. 1987; Sivaselvan and Reinhorn 2001), which is obtained by extending the 153 initial elastic slope backwards until it reaches a force of αFhy in the third coordinate plane. N 0.5 0.5sign F .du h (5) 154 It is observed in Figure 5 that if a load reversal occurs at a force Fc and displacement uc, 155 the degraded hysteretic elastic stiffness would be khd, which is the slope of the line joining the 156 load reversal point (Fc, uc) and the pivot point. The pivot point is obtained by extending the line 157 with initial stiffness on the reverse side until it reaches a force of –(αFhy). 158 degradation factor Rk is applied to the initial elastic stiffness kho to obtain the degraded stiffness 159 khd. The factor can be obtained from geometry as: 160 Rk Fc Fhy / khouc Fhy The stiffness (6) 161 Stiffness degradation is incorporated into the overall hysteretic model by multiplying the 162 hysteretic stiffness by Rk. This ensures that stiffness degradation is applied to the hysteretic 9 163 component of the total tangent stiffness. Although Erlicher and Bursi (2008) recently reported 164 some drawbacks of the pivot-point rule, it appears to be acceptable for the macro analysis of 165 structural elements. 166 Strength Degradation: A change (or deterioration) of the initial yield strength of the hysteretic 167 component Fhyo is defined as strength degradation. Such degradation can be a result of loss of 168 strength of sections or elements at large deformations (or large ductility demands), and due to 169 unrecoverable energy (H) dissipating through cyclic behavior. The degraded yield force (Fhy) can 170 therefore be adjusted by a ductility based factor and an energy based factor according to 171 Sivaselvan and Reinhorn (2001): 172 Fhy Fhyo 1 umax ucap 173 The expression in the first parenthesis of Equation (7) is the ductility based strength degradation 174 factor, where |umax|/ucap is the displacement ductility based on maximum displacement capacity 175 ucap, while umax is the latest maximum displacement attained; and β1 is the ductility based 176 strength degradation parameter. The expression in the second parenthesis is the energy based 177 strength degradation factor, where β2 is the energy based strength degradation parameter. Such 178 degradation was observed in experiments and was quantified by numerous researchers. Hult is the 179 ultimate unrecoverable work, i.e., energy dissipated if the element is forced by a monotonic 180 displacement until it reaches ucap. Note that uy<|umax|<ucap. The functions H and Hult are 181 calculated as follows: 182 H Fh .du Fh 2 2 Rk kh 1/ 1 1 u 0 and 2 1 2 H H ult H ult Fh you y / 2 Fhyo ucap u y (7) (8) 10 183 Strength degradation can be incorporated in the global hysteretic model using modified 184 Fhy from Equations (7) and (8) to calculate the tangent stiffness ktangent in Equation (5). 185 Bond slip and “pinched” hysteretic formulation: 186 displacements resulting from loss of bond of section components, or opening and closing of 187 cracks, among others. These phenomena lead to engagement and disengagement of the main 188 elastic characteristics of the section. Pinching or slip can be incorporated in the model by 189 considering an additional stiffness kslip in series with the hysteretic stiffness. 190 Gaussian “pinching” model: An expression for kslip developed according to a Gaussian model 191 proposed by Baber and Noori (1985) and Reinhorn et al. (1995) is re-expressed here as follows: 192 kslip 1/2 Fh / Fhy Z 1/ 2 Rs umax umax H umax umax e Pinching and slip are changes in Zs 2 Z s Fhy 1 (9) 193 , umax Here Fh is the current hysteretic force and Fhy is the yield force of the hysteretic spring. umax 194 are the maximum positive and negative displacements, respectively. Zs is a parameter denoting 195 the sharpness of the slip. Parameter Rs controls the slip-length and Z is the slip-center, mostly 196 taken as zero. The above model is suitable for elements with smooth pinching (Zs>0.05) and 197 smaller slip-length factors (Rs<0.5). In cases where elements have sharper pinching (Zs<0.01) 198 and large slip-length (Rs>0.8), the above model fails to produce the desired hysteresis loop. In 199 addition, due to numerical limitations in computing devices, the exponential term in Equation (9) 200 may reach a null value for lower Zs and the solution may stop. Moreover, experimental data for 201 some structural elements suggest that Rs should not remain constant, and in fact it should 202 increase at larger deformations. An improved model is developed in the following sections. 203 11 204 205 206 207 208 209 210 211 212 Hysteretic model with asymmetric yielding properties: Some structural elements such as those made of steel or concrete, may have a single elastic stiffness ko, but may yield at different levels in the load reversals. In this case, instead of a single Fhyo, there will be two different F+hyo and Fhyo, associated with two different u+yand u-y. The formulation suggested here is based on the fact that a “zero” force axis can be defined at the median of the yielding forces F+hyo and F-hyo, and the behavior is then almost identical to a system with symmetric yielding. This can be achieved with a translational transformation of the zero axis at each step of the numerical computation. Therefore, while preserving the principle of kinematic hardening, the case of a system with different positive and negative yield forces can be reformulated as follows: 213 a) Equation (7), used to calculate degraded yield strength, is modified as: 214 1/ 1 / / / / 1 2 1 2 H H ult Fhy Fhyo 1 u max ucap F F 1 sign(du ) / 2 F 1 sign(du ) / 2 hy hy hy 215 b) Calculation of the average Fhy and uy and modification of Eq. (8) to calculate Hult is modified 216 as: 217 F F F / 2, u u u / 2 y av y y hy av hyo hyo H ult Fhy av . u y av / 2 Fhy av ucap u y av 218 c) Equation (6), used to calculate the stiffness degradation factor, is modified as: 219 R / F + F / / k u F / , c yo o c k yo Rk Rk 1 sign(du ) / 2 Rk 1 sign(du ) / 2 220 It may be observed that Equation sets (10) and (12) are formulated such that Fhyo and Fhy will be 221 used in the calculations when the load-displacement path approaches a positive yield line, while (10) (11) (12) 12 222 Fhyo and Fhy will be used when it approaches a negative yield line. In addition, these formulations 223 offer the flexibility of choosing different degradation parameters, , 1 and 2 , for positive and 224 negative loading. In Equation set (11), the average of the positive and negative yield forces and 225 positive and negative displacements are used to calculate Hult. 226 Extended Smooth Hysteretic Model 227 The current re-formulation of the smooth hysteretic model of Sivaselvan and Reinhorn (2001) is 228 extended herein by including the aforementioned features as follows. 229 Smooth Hysteretic Model with Nonlinear Hardening: Hysteresis with nonlinear elastic 230 behavior is modeled considering two springs acting in parallel. The first is the Hysteretic Spring 231 with yield force Fhy, yield displacement uy and elastic hysteretic stiffness kh. The additional 232 spring is the Nonlinear Elastic Spring, whose stiffness, k(u), depends upon the displacement u. 233 k(u) can be expressed as the summation of a constant linear elastic stiffness kle, and a nonlinear 234 elastic stiffness component kne.f<𝑢̅>, which is discretely applied at 𝑢̅. 𝑢̅ is a displacement 235 variable normalized by the a gap length, while kne is a dimensional stiffness coefficient; and 236 <x> is the Macaulay bracket defined as <x>=x for x>0, and =0 otherwise. 237 <x> is expressed computationally as (x+|x|)/2 and f is a function that defines the nonlinearity of 238 k(u) with respect to u. In this formulation, Hence, k(u) is given by: 239 k (u) kle kne f u 240 The displacement dependent part of k(u) is activated after the absolute deformation |u| is 241 beyond a specified “gap” length ug or another shifting displacement. This can be generalized by 242 expressing the normalized displacement variable 𝑢̅ as a ratio of |𝑢|and ug, and defining f<𝑢̅ > as: (13) 13 243 244 f u u ug u u g or u / ug 1 u /u g 1 (14) The instantaneous tangent stiffness ktangent of the two springs taken together is given by: 0.5 0.5sign F .du k (u ) 245 ktan gent kh 1 Fh Fhy 246 Fh is the current force and Fhy is the yield force strength of the hysteretic spring. Note that in the 247 presence of the initial hysteretic stiffness kh, the linear elastic component kle of the overall initial 248 stiffness ko is given by: 249 kle ko kh 250 In addition, the total apparent yield force Fy of the overall behavior is given by: N (15) h (16) uy 251 Fy Fhy k (u ).du (17) 0 252 Embedding Strength Degradation in the Instantaneous Stiffness: SHM can accommodate 253 strength degradation in its hysteretic part. An instantaneous tangent-stiffness of a deforming 254 section in a structural element can be obtained by differentiating both sides of Equation (2) with 255 respect to u as: 256 dF du a. Fy u y (1 a) Fy dZ du (1 a)Z dFy du ktan gent a. Fy u y (1 a) Fy u y 1 Z N or 0.5sign(Z .du) 0.5 (1 a)Z dFy (18) du 257 Note that if Fy changes, which can happen during strength degradation, the last term in Equation 258 (18) comprising dFy/du defines the rate of variation (shrinking or expanding) of the yield surface. 259 Stiffness Degradation of Linear Elastic Spring: Experimental results often suggest that in 260 addition to degradation of elastic stiffness of the hysteretic spring kh, the linear elastic stiffness 261 kle of the nonlinear elastic spring k(u) also degrades, depending upon the displacement ductility. 14 262 Hence, a one parameter degradation model, based on maximum displacement umax, is proposed 263 for kle: 264 kl e deg raded kle initial 1 1 umax 265 (kle)degraded and (kle)initial are the degraded and initial stiffness of the elastic spring, respectively. η1 266 is a positive factor that is zero for non-degrading kle, while ucap is the maximum displacement 267 capacity. 268 Modified Gaussian Pinching Model: In order to overcome the previously discussed numerical 269 problems, the exponent in Equation (9), 1/2 Fh / Fhy Z 270 term stays within the limitations of the computing device and it allows a smooth formulation. 271 The slip length factor Rs is varied linearly according to the maximum achieved displacement as: 272 Rs Rso Rs max Rso 2 umax / ucap 273 Rso is the initial value of Rs, Rsmax is the maximum value of Rs, η2 is a positive factor, umax is the 274 maximum displacement, and ucap is the displacement capacity. For Rs to be constant η2=0. 275 Alternate Pinching Model based on Transcendental Function: As an alternative to the 276 Gaussian Pinching Model, a pinching model based on a Tangent Function is formulated. This 277 automatically alleviates the numerical difficulties faced by the Gaussian formulation and 278 provides more control of the slip stiffness kslip, especially for degradations. In this model, the 279 forcing function of the slip-lock spring is defined by: 280 F 281 Fslip and uslip are the force and displacement of the slip-lock spring, respectively, as shown in 282 Figure 7. As the slip-lock spring and hysteretic springs are connected in series, they essentially slip / ucap (19) Zs 2 , is “capped” such that the exponential Fs tan uslip 2us (20) (21) 15 283 share the same force and thus, Fhys=Fslip. Fs is the crack-closing force associated with the bond- 284 slip. It is expressed as a fraction of the maximum hysteretic force Fhy as: 285 Fs s1Fhy 286 us is the slip (displacement) and is equal to the permanent deformation: 287 us umax u y 288 Due to uncertainties in the crack opening and in the permanent deformations, an adjustment 289 factor s2 is introduced to the right side of Equation (23). In this case, the slip us is defined as: 290 us s2 umax u y 291 The tangent stiffness of the slip-lock spring is obtained from the derivative of Equation (21) with 292 respect to uslip : 0<s1 1.0 (22) (23) 0<s 2 1.0 (24) kslip dFslip duslip Fs us / 2 sec 2 uslip 2us 293 or k slip kslip min 1 Fslip / Fs 294 where kslipmin is the minimum stiffness of the slip-lock spring and is defined as: 295 kslip min Fs us / 2 296 The slip-lock spring is attached in series with the hysteretic spring and their combined stiffness 297 (kcomb ) is given by: 298 kcomb khys .kslip 299 k hys k slip 2 (25) (26) (27) 16 300 Examples of Applications of Smooth Hysteretic Models 301 The above formulations were implemented in MATLABTM and further integrated in the 302 computational package IDARC2D by Reinhorn et al. (2009). Some examples that illustrate the 303 sensitivity of the model to the specified parameters are presented below. 304 Example 1: 305 experiencing large lateral deformations require the influence of geometric nonlinearity to be 306 incorporated. 307 The 308 250, kle=20.0,N=20,kne=100;𝑢𝑐𝑎𝑝 = 25( 𝑢𝑦+ + 𝑢𝑦− ), 𝑢𝑔 = 0.75 309 parameters are considered: mild, moderate, and severe (Sivaselvan and Reinhorn, 1999). No 310 degradation of kle is considered. Stiffening at Large Displacements: Typical models of bending elements Assume first a simple structural element subjected to sinusoidal displacement. properties of the element are: kh = 980, Three + − 𝐹ℎ𝑦𝑜 = 500, 𝐹ℎ𝑦𝑜 = sets of degradation 311 In the first set of analyses, slip-pinching at large deformation is obtained using 312 parameters Rso = 0.25, η2=0, Zs = 0.10, 𝑍̅ = 0.0, according to the Gaussian Model. In the second 313 analysis set (random displacement), all other primary variables except ug are kept the same as in 314 the first set of analyses. The “gap” displacement ug is assumed to be 0.5 in. for this analysis. 315 Also, the effect of pinching is discarded in the second set of analyses. The force-displacement 316 responses and the displacement history chosen for the above analyses are shown in Figure 8. 317 The sinusoidal displacement analysis demonstrates the effect of stiffening at large 318 displacement, coupled with pinching and degradation of strength and stiffness. Adherence of the 319 model to the kinematic hardening principle is also visible in this analysis. The random 320 displacement analysis shows the capability of the model to capture the effect of strength and 321 stiffness degradation without any instability. 17 322 Example 2: Softening at Large Displacement: Many structural elements show nonlinear 323 hysteretic behavior with softening at larger deformations. The foregoing formulations are 324 implemented in MATLABTM to simulate a lateral force deflection diagram for a beam-column 325 moment connection (Specimen UCSD-1R from the SAC Joint Venture 1996, Sivaselvan and 326 Reinhorn 2001). The following parameters are used in the modeling: Fyho =120 kips; kho = 120 327 kip/inch; kle = 10 kip/inch; kne = -20 kip/inch; ug = 2.5 inch; N=2.0. The strength and stiffness 328 degradation are modeled using the following parameters: α = 5.0; β1 = 0.4; β2=0.2; η1=0.0; ucap = 329 20.0uy. A comparison between the experimental and analytical responses is shown in Figure 9. 330 Example 3 : Hysteretic Behavior with Geometric Nonlinearity: Laterally loaded elements, such 331 as flexible columns or partition wall structures, may exhibit large deformations when integrated 332 in ductile systems. At such large deformations, the pure bending and transverse shear 333 deformations are coupled with the axial deformations, as shown in Figure 10. However, the 334 large deformations may recover upon release of loading and the axial coupling behavior remains 335 elastic. Thus, the entire lateral force displacement behavior becomes nonlinear. The tangent 336 stiffness ktangent of this nonlinear behavior will have contributions from flexural stiffness kf and 337 shear stiffness ks (both of which are hysteretic), and also from the elastic axial stiffness kao. 338 Formulation of tangent stiffness: 339 stiffness, ktangent, depend upon their respective bilinear hysteresis and can be determined as 340 described for infill walls by Reinhorn et al. (2009). In this example, the lateral stiffness, 341 irrespective of the contributions from either or both kf and ks, is called klateral. The nature and thus the value of kf and ks in the tangent 342 The laterally loaded member of length L shown in Figure 10 displaces from point A to 343 A’, by a distance u, through the action of lateral load F. The member is fixed at point O. At A’, 344 the slope of the member is . The lateral force F can be resolved into component Fs, normal to 18 345 the deflected shape at A’, and parallel to the deflected shape at Fa. The components of total 346 displacement u are us along Fs, and ua along Fa. The transverse force Fs is resisted by the 347 stiffness klateral and the axial force Fa by kao. The displacement components are: 348 ua u sin ; us u cos 349 After resolving F into components, the total force is expressed as: 350 F Fa sin Fs cos kaou sin 2 klateral u cos 2 351 When u is small compared to L, u / L, sin u / L , cos 1.0 . Using these 352 approximations, Equation (29) is modified to: 353 (28) F kao u 3 / L2 klateral u (29) (30) 354 The derivative of both sides of Equation (30) produces the instantaneous (tangent) stiffness: 355 dF du k tan gent klateral 3kao u / L 356 Note that in Equation (31), klateral is hysteretic with finite post-elastic stiffness and the term 357 containing kao is nonlinear elastic. Assuming the hysteretic part of klateral to be khel and the post- 358 elastic lateral stiffness to be kpel, coupling of axial and lateral stiffness can be incorporated in the 359 present hysteretic model by defining the primary parameters of the model as follows: 360 kh khel , 361 f u 3 u / L kel k pel , 2 ken kao (31) (32) 2 (33) 362 If the physical situation demands that the coupling of axial stiffness occurs after the displacement 363 u exceeds a certain gap length ug, then Equation (33) can be modified as: 364 f u 3 u ug L 2 (34) 19 365 Implementations and Sample Results: A model of a partition wall element, subjected to 366 sinusoidal quasi-static displacements applied at the top of the wall, was tested in University at 367 Buffalo (SUNY) by Filiatrault et al. (2010). It is used here to illustrate the performance of the 368 hysteretic model with large deformations. 369 From the experimental Base-Shear vs. Top Displacement hysteresis shown in Figure 370 11(a), it is observed that the gap length ug is a varying parameter dependent on the maximum slip 371 umax reached before the wall collapsed. The variation can be expressed as: 372 ug ugo 1 umax ugo 373 ugo is the initial gap length, g1 is a factor (<1) and umax is the absolute maximum displacement 374 reached in the last cycle. The gap length was measured by Filiatrault et al. (2010) during the 375 experiments, when the tested partition wall slipped at both the top and bottom, depending on the 376 maximum displacements. It was also observed that while using the Alternative Pinching Model 377 based on the tangent function, the parameters s1 and s2 should also be varied because Fs and us 378 change nonlinearly with umax at large displacements. The variation of s1 and s2 can be expressed 379 as: 380 s1 s1o 1 s11 umax u _ cap s1min s2 s2 o 1 s22 umax u _ cap s2max 381 s1o and s2o are the initial values and s1max and s2max are the maximum possible values of s1 and s2. 382 The properties of the partition wall estimated for the Gaussian Pinching Model are shown in 383 Table 1 (left). A comparison between the experimental and analytical responses for the Modified 384 Gaussian Pinching Model is shown in Figure 11(b). u max u go g1umax umax u go u max u go (35) (36) 20 385 The parameters of the analysis performed using the new formulation based on the tangent 386 function for kslip and Equation (35) for gap length are shown in Table 1 (right). The 387 corresponding comparison between the experimental and analytical responses is shown in Figure 388 11(c). 389 directly from the experimental results. All the other input parameters were determined from 390 material data. Since these experiments were displacement driven, the maximum force levels of 391 the numerical simulations at different displacements are compared with those of the experimental 392 results. It is found that the analytical force magnitudes were within 10% of the experimental 393 values. If the nonlinear parameters are not rate dependent (such as velocity dependent damping), 394 the selected parameters produce reasonable results in dynamic analyses, within the ductility 395 range considered. 396 Reinhorn (2001), and the comparison of the simulation with the experimental results is shown in 397 Figure 12. It is observed that the previous SHM, which uses the Gaussian pinching model, 398 shows unstable behavior for sharper slip (pinching) and at large displacements. It also fails to 399 explicitly capture the degradation of nonlinear elastic stiffness and stiffening at large 400 deformations. 401 Remarks and Conclusions 402 The models presented in this paper account for the basic re-formulations of the SHM that are 403 time-independent, as opposed to the more classical Bouc–Wen model, which seems to be 404 dependent on force and displacement rates (velocities). Therefore, the model presented herein 405 can be used in nonlinear static analysis as well as in dynamic analyses, to consider degradations 406 and deteriorations. It is also observed that the Alternative Pinching Formulation based on a Note that the initial stiffness and yield force in the last two examples were obtained This example is also simulated using the previous SHM by Sivaselvan and 21 407 tangent function offers more flexibility and control for “pinched” modeling, especially when 408 compared to the more classical Gaussian Pinching Model. 409 The stiffness properties of nonlinear structural elements can be formulated using 410 information based on the material and geometry of sections derived from elementary nominal 411 formulations; however, for degradation and pinching, a database must be developed, as was done 412 in the experimental study developed by ASCE/SEI (2006) and Filiatrault et al. (2010). Kunnath 413 et al. (1997) also carried out system identification of degrading and pinched reinforced concrete 414 structures in order to calibrate the extended Bouc-Wen SHM (Kunnath et al. 1990). 415 The stability of the model is related to the continuous positive-definite stiffness matrix. 416 The strategy offered here is to use this positive stiffness, while loss of strength due to 417 degradation is coupled to the solution of the global system. There are, however, three possible 418 conditions that can lead to instability in the SHM: a) the hysteretic stiffness kho is much larger 419 (more than 100 times) than the linear elastic stiffness kle, b) the exponent of transition from the 420 elastic to the post-elastic state is very large (N larger than 100), and c) the "pinching" exponential 421 term is beyond the capacity of the computing device in the previous Gaussian pinching model 422 (Baber and Noori 1985). While the first two cases can be handled by proper selection of the 423 model parameters and by reduction of computational step size, the third case is addressed herein 424 by modifying the Gaussian model. 425 The enhancements to the proposed SHM are incremental to the previous version 426 developed by Sivaselvan and Reinhorn (2001). The use of the previous version without the 427 improvements remains valid when suitably fitted coefficients are used. The new model was 428 integrated in the computational platform IDARC2D (Reinhorn et al. 2009) and was found to be 22 429 backward compatible and able to successfully simulate all the examples included in the software 430 validation package simulated with PHM and the previous version of SHM. 431 The one-dimensional tangent stiffness calculated by the proposed models can be further 432 incorporated into the global stiffness matrix in order to analyze the structure as a whole. The 433 model with geometric nonlinearity presented herein may be further modified to incorporate the 434 effect of axial load on the element as discussed in Kikuchi et al. (2007). The model may also be 435 extended to bi-axial force-deformation (or moment-curvature) relationship required for three- 436 dimensional analysis as described in Sivaselvan and Reinhorn (2004). Finally, the one- 437 dimensional model developed herein can serve as a base for three-dimensional models developed 438 by Ray (2012). 439 23 440 References 441 442 06 ASCE/SEI (2006). "Seismic Rehabilitation of Existing Buildings (41-06) " ASCE/SEI 41- 443 444 Baber, T. T., and Noori, M. N. (1985). "Random Vibrations of Degrading Pinching Systems." Journal of Engineering Mechanics, 111(8), 1010-1026. 445 446 Bouc, R. (1967). "Forced Vibration of Mechanical System with Hysteresis." Proc., 4-th Conf. on Nonlinear Oscillations. 447 448 449 Casciati (1987). "Nonlinear Stochastic Dynamics of Large Structural Systems By Equivalent Linearization." Proc., ICASP5, Fifth Internation Conference on Application of Statistics and Probability in Soil and Structural Engineering. 450 451 Chen, Y. F., and Han, D. J. (2007). "Plasticity for Structural Engineers." J. Ross Publishing, Inc., Fort Lauderdale, 14-16. 452 453 454 Clough, R. W., and Johnston, S. B. (1966). "Effects of Stiffness Degradation on Earthquake Ductility Requirements." Proc., Second Japan Earthquake Engineering Symposium, 227-232. 455 456 Constantinou, M. C., and Adnane, M. A. (1987). "Dynamics of Soil-Base-IsolatedStructure Systems." National Science Foundation. 457 458 459 Erlicher, S., and Bursi, O. S. (2008). "Bouc-Wen-Type Models with Stiffness Degradation: Thermodynamic Analysis and Applications." ASCE Journal of Engineering Mechanics, 134(10), 843-855. 460 461 462 Filiatrault, Andre, Mosqueda, Gilberto, Retamales, Rodrigo, Davies, Ryan, Tian, Yuan, and Fuchs, Jessica (2010). "Experimental Seismic Fragility of Steel Studded Gypsum Partition Walls and Fire Sprinkler Piping Subsystems." ASCE, 237-237. 463 464 465 Ibarra, L. F., Medina, R. A., and Krawinkler, H. (2005). "Hysteretic models that incorporate strength and stiffness deterioration." Earthquake Engineering and Structural Dynamics, 34, 1489-1511. 466 467 Iwan, W. D. (1966). "A Distributed Element Model for Hysteresis and its Steady-State Dynamic Response." Journal of Applied Mechanics, ASME, 33(4), 893-900. 468 469 470 Kikuchi, M., Yamamoto, S., and Aiken, I. D. (2007). "An Analytical Modeling of LeadRubber Bearings under Large Deformation." 8th Pacific Conference on Earthquake EngineeringSingapore. 471 472 Kunnath, S., Reinhorn, A., and Park, Y. (1990). "Analytical Modeling of Inelastic Seismic Response of R/C Structures." Journal of Structural Engineering, 116(4), 996-1017. 24 473 474 475 Kunnath, Sashi K., Mander, John B., and Fang, Lee (1997). "Parameter identification for degrading and pinched hysteretic structural concrete systems." Engineering Structures, 19(3), 224-232. 476 477 478 Kunnath, S. K., and Reinhorn, A. M. (1994). "Efficient Modeling Scheme for Transient Analysis of Inelastic RC Structures." Microcomputers in Civil Engineering, International Journal of Computer-Aided Civil and Infrastructure Engineering, 10, 97-100. 479 480 481 Nagarajaiah, Satish, Reinhorn, Andrei M., and Constantinou, M. C. (1991a). "Nonlinear Dynamic Analysis of 3-D-Base-Isolated Structures." ASCE/Journal of Structural Engineering, 117(7), 2035-2054. 482 483 484 Nagarajaiah, Satish, Reinhorn, Andrei M., and Constantinou, M. C. (1991b). "3D-BASIS: Nonlinear dynamic analysis of three-dimensional base isolated structures - Part II." NCEER-910005, National Center for Earthquake Engineering Research, SUNY, Buffalo, New York 485 486 487 Park, Y. J., Reinhorn, A. M., and Kunnath, S. K. (1987). "IDARC:Inelastic Damage Analysis of Reinforced Concrete Frame - Shear-Wall Structures." Technical Report NCEER, University at Buffalo, SUNY. 488 489 490 Reinhorn, A. M., Madan, A., Valles, R. E., Reichmann, Y., and Mander, J. B. (1995). "Modeling of Masonary Infill Panels for Structural Analysis." Technical Report NCEER, University at Buffalo, SUNY. 491 492 493 Reinhorn, A. M., Roh, H. S., Sivaselvan, M. V., Kunnath, S. K., Valles, R. E., Madan, A., Li, C., Lobo, R., and Park, Y. J. (2009). "IDARC2D Version 7.0: A Program for the Inelastic Damage Analysis of Structures." Technical Report MCEER, University at Buffalo, SUNY. 494 495 Rosenbrock, H. H. (1963). "Some general implicit processes for the numerical solution of differential equations." The Computer Journal, 329-330. 496 497 Sivaselvan, M., and Reinhorn, A. M. (2001). "Hysteretic Models for Deteriorating Inelastic Structures." ASCE/Journal of Engineering Mechanics, 126(6), 633-640. 498 499 500 Sivaselvan, M. V., and Reinhorn, A. M. (2004). "Nonlinear Structural Analysis Towards Collapse Simulation : A Dynamical Systems Approach " MCEER-04-0005, University at Buffalo, SUNY, Buffalo. 501 502 Sivaselvan, M.V., and Reinhorn, A.M. (1999). "Hysteretic Models for Cyclic Behavior of Deteriorating Inelastic Structures." Technical Report MCEER, University at Buffalo, SUNY. 503 504 505 Takeda, T., Sozen, M. A., and Neilsen, N. N. (1970). "Reinforced Concrete Response to Simulated Earthquake." Journal of Structural Division, Proceedings of the American Society of Civil Engineers, 96(12), 2557-2573. 506 507 Thyagarajan, R. S. (1989). "Modeling and Analysis of Hysteretic Structural Behavior." PhD, California Institute of Technology, Pasadena. 25 508 509 Wen, Y. K. (1976). "Methos for Random Vibration of Hysteretic Systems." J. Engrg. Mech. Div., ASCE, 102(2), 249-263. 510 511 Ray, T. (2012). "Modeling of Multidimensional Nonlinear Elastic and Inelastic Structural Systems."PhD, University at Buffalo, SUNY. 512 513 514 515 Rodgers, Geoffrey W., Mander, John B., and Geoffrey Chase, J. (2012). "Modeling cyclic loading behavior of jointed precast concrete connections including effects of friction, tendon yielding and dampers." Earthquake Engineering & Structural Dynamics, 41(15), 22152233. 516 517
© Copyright 2026 Paperzz