1D elastic full‐waveform inversion and uncertainty estimation by means of a hybrid genetic algorithm–Gibbs sampler approach

Stochastic optimization methods, such as genetic algorithms, search for the global minimum of the misfit function within a given parameter range and do not require any calculation of the gradients of the misfit surfaces. More importantly, these methods collect a series of models and associated likelihoods that can be used to estimate the posterior probability distribution. However, because genetic algorithms are not a Markov chain Monte Carlo method, the direct use of the genetic‐algorithm‐sampled models and their associated likelihoods produce a biased estimation of the posterior probability distribution. In contrast, Markov chain Monte Carlo methods, such as the Metropolis–Hastings and Gibbs sampler, provide accurate posterior probability distributions but at considerable computational cost. In this paper, we use a hybrid method that combines the speed of a genetic algorithm to find an optimal solution and the accuracy of a Gibbs sampler to obtain a reliable estimation of the posterior probability distributions. First, we test this method on an analytical function and show that the genetic algorithm method cannot recover the true probability distributions and that it tends to underestimate the true uncertainties. Conversely, combining the genetic algorithm optimization with a Gibbs sampler step enables us to recover the true posterior probability distributions. Then, we demonstrate the applicability of this hybrid method by performing one‐dimensional elastic full‐waveform inversions on synthetic and field data. We also discuss how an appropriate genetic algorithm implementation is essential to attenuate the “genetic drift” effect and to maximize the exploration of the model space. In fact, a wide and efficient exploration of the model space is important not only to avoid entrapment in local minima during the genetic algorithm optimization but also to ensure a reliable estimation of the posterior probability distributions in the subsequent Gibbs sampler step.     

Spatial behaviour of Rayleigh waves in layered half‐spaces under active surface sources

In the free state, Rayleigh waves are assumed to travel in the form of planar wavefronts. Under such an assumption, the propagation behaviour of the modes of Rayleigh waves in layered half‐spaces is only frequency dependent. The frequency behaviour, which is often termed as dispersion, is determined by the shear wave velocity profile of layered soils within the depth related to wavelength (or frequency). According to this characteristic, the shear wave velocity profile can be back‐analysed from the dispersion. The technique is widely used in the surface wave testing. However, the wavefronts of Rayleigh waves activated by the surface sources are non‐planar. The geometric discrepancy could result in Rayleigh waves manifesting distance‐dependent behaviour, which is referred to as spatial behaviour in this paper. Conventional analysis ignoring this spatial behaviour could introduce unexpected errors. In order to take the effects of sources on the propagation behaviour into account, a new mathematical model is established for Rayleigh waves in layered elastic media under vertical disc‐like surface sources using the thin‐layer method. The spatial behaviour of the activated modes and the apparent phase velocity, which is the propagation velocity of Rayleigh waves superposed by the multiple modes, are then analysed. Aspects of the spatial behaviour investigated in this paper include the equilibrium path, the particle orbit, and the geometric attenuation of the activated Rayleigh waves. The results presented in this paper can provide some guidelines for developing new inverse mathematical models and algorithms.     

Influence of earthquake ground‐motion duration on damage estimation: application to steel moment resisting frames

This paper presents an analytical study evaluating the influence of ground motion duration on structural damage of 3‐story, 9‐story, and 20‐story SAC steel moment resisting frame buildings designed for downtown Seattle, WA, USA, using pre‐Northridge codes. Two‐dimensional nonlinear finite element models of the buildings are used to estimate the damage induced by the ground motions. A set of 44 ground motions is used to study the combined effect of spectral acceleration and ground motion significant duration on drift and damage measures. In addition, 10 spectrally equivalent short‐duration shallow crustal ground motions and long‐duration subduction zone records are selected to isolate duration effect and assess its effect on the response. For each ground motion pair, incremental dynamic analyses are performed at at least 20 intensity levels and response measures such as peak interstory drift ratio and energy dissipated are tracked. These response measures are combined into two damage metrics that account for the ductility and energy dissipation. Results indicate that the duration of the ground motion influences, above all, the combined damage measures, although some effect on drift‐based response measures is also observed for larger levels of drift. These results indicate that because the current assessment methodologies do not capture the effects of ground motion duration, both performance‐based and code‐based assessment methodologies should be revised to consider damage measures that are sensitive to duration. Copyright © 2016 John Wiley & Sons, Ltd     

Experimental study of deformable connection consisting of buckling‐restrained brace and rubber bearings to connect floor system to lateral force resisting system

This paper presents experimental and numerical studies of a full‐scale deformable connection used to connect the floor system of the flexible gravity load resisting system to the stiff lateral force resisting system (LFRS) of an earthquake‐resistant building. The purpose of the deformable connection is to limit the earthquake‐induced horizontal inertia force transferred from the floor system to the LFRS and, thereby, to reduce the horizontal floor accelerations and the forces in the LFRS. The deformable connection that was studied consists of a buckling‐restrained brace (BRB) and steel‐reinforced laminated low‐damping rubber bearings (RB). The test results show that the force–deformation responses of the connection are stable, and the dynamic force responses are larger than the quasi‐static force responses. The BRB+RB force–deformation response depends mainly on the BRB response. A detailed discussion of the BRB experimental force–deformation response is presented. The experimental results show that the maximum plastic deformation range controls the isotropic hardening of the BRB. The hardened BRB force–deformation responses are used to calculate the overstrength adjustment factors. Details and limitations of a validated, accurate model for the connection force–deformation response are presented. Numerical simulation results for a 12‐story reinforced concrete wall building with deformable connections show the effects of including the RB in the deformable connection and the effect of modeling the BRB isotropic hardening on the building seismic response. Copyright © 2016 John Wiley & Sons, Ltd.     

Nonsmooth dynamics prediction of measured bridge response involving deck‐abutment pounding

Earthquake‐induced deck‐abutment contact alters the boundary conditions at the deck level and might activate a different mechanical system than the one assumed during the design of the bridge. Occasionally this discrepancy between the assumed and the actual seismic behavior has detrimental consequences, for example, pier damage, deck unseating, or even collapse. Recently, an insightful shake‐table testing of a scaled deck‐abutment bridge model 1 , showed unexpected in‐plane rotations even though the deck was straight. These contact‐induced rotations produced significant residual displacements and damage to the piers and the bents. The present paper utilizes that experimental data to examine the validity and the limitations of a proposed nonsmooth dynamic analysis framework. The results show that the proposed approach satisfactorily captures the planar rigid‐body dynamics of the deck which is characterized by deck‐abutment contact. The analysis brings forward the role of friction on the physical mechanism behind the rotation of the deck, and underlines the importance of considering the frictional contact forces during deck‐abutment interaction even for straight bridges, which typically are neglected. Finally, the paper investigates the sensitivity of the rotation with respect to macroscopic contact parameters (i.e., the coefficient of friction and the coefficient of restitution). Copyright © 2017 John Wiley & Sons, Ltd.     

Determination of the seismic performance factors for post‐tensioned rocking timber wall systems

Post‐tensioned technologies for concrete seismic resistant buildings were first developed in the 1990s during the PREcast Seismic Structural Systems program. Among different solutions, the hybrid system proved to be the most resilient solution providing a combination of re‐centering and energy dissipative contributions respectively by using post‐tensioned tendons and mild steel reinforcement. The system, while providing significant strength and energy dissipation, reduces structural element damage and limits post‐earthquake residual displacements. More recently, the technology was extended to laminated veneer lumber (LVL) structural members, and extensive experimental and numerical work was carried out and allowed the development of reliable analytical and numerical models as well as design guidelines. On the basis of the experimental and numerical outcomes, this paper presents the evaluation of the seismic performance factors for post‐tensioned rocking LVL walls using the FEMA P‐695 procedure. Several archetype buildings were designed considering different parameters such as the building and story height, the type of seismic resistant system, the magnitude of gravity loads and the seismic design category. Lumped plasticity models were developed for each index archetype to simulate the behavioral aspects and collapse mechanisms. Non‐linear quasi‐static analyses were carried out to evaluate the system over‐strength factor; moreover, non‐linear time history analyses were performed using the incremental dynamic analysis concept to assess the collapse of each building. From the results of quasi‐static and dynamic analyses the response modification factor, R, system over‐strength factor, Ω₀, and deflection amplification factor, C_d, values of, respectively, 7, 3.5 and 7.5 are recommended. Copyright © 2016 John Wiley & Sons, Ltd.     

Cyclic behavior and failure mechanism of self‐centering energy dissipation braces with pre‐pressed combination disc springs

A new type of bracing system composed of friction energy dissipation devices for energy dissipation, pre‐pressed combination disc springs for self‐centering and tube members as guiding elements is developed and experimentally studied in this paper. The mechanics of this system are explained, the equations governing its hysteretic responses are outlined and large‐scale validation tests of two braces with different types of disc springs are conducted under the condition of low cyclic reversed loading. The experimental results demonstrate that the proposed bracing system exhibits a stable and repeatable flag‐shaped hysteretic response with an excellent self‐centering capability and effective energy dissipation throughout the loading protocol. Furthermore, the maximum bearing force and stiffness are predicted well by the equations governing its mechanical behavior. Fatigue and destructive test results demonstrate that the proposed bracing system can maintain stable energy dissipation and self‐centering capabilities under large deformation cyclic loading even when the tube members exceed the elastic limit and that a larger bearing capacity is achieved by the system that has disc springs without a bearing surface. Copyright © 2016 John Wiley & Sons, Ltd.     

Residual slip of sliding blocks induced by near‐fault ground motions

The response of a rigid block supported on a horizontally moving foundation through a dry‐friction contact is investigated to near‐fault ground motions. Such motions can be thought of as consisting of a coherent component (‘pulse’) and an incoherent component, which can be described as a band‐limited ‘random noise’. The equation of motion of this strongly nonlinear system is reduced to a normalized form that reveals important parameters of the problem such as the critical acceleration ratio. The response of the sliding block to a set of uniformly processed near‐fault motions, covering a sufficiently wide range of magnitudes, is evaluated numerically for selected discrete values of the acceleration ratio. For each value of the critical acceleration ratio, the numerically computed residual slips are fitted with a Weibull (Gumbel type III) extreme value probability distribution. This allows the establishment of regression equations that describe accurately design sliding curves corresponding to various levels of non‐exceedance probability. The analysis reveals that the coherent component of motion contributes significantly to the response of the sliding block. Furthermore, the relevant acceleration in specifying the critical acceleration ratio is the (normalized) amplitude, α_{H_pulse}, of the pulse and not the (normalized) amplitude of the incoherent component α_H. Finally, the incoherent component is described quantitatively in terms of the root‐mean‐square acceleration a_RMS, and an attempt is made to understand its influence on the response of the sliding block. Copyright © 2016 John Wiley & Sons, Ltd.     

Probabilistic analysis of soil‐structure interaction effects on the seismic performance of structures

This paper revisits the phenomenon of dynamic soil‐structure interaction (SSI) with a probabilistic approach. For this purpose, a twofold objective is pursued. First, the effect of SSI on inelastic response of the structure is studied considering the prevailing uncertainties. Second, the consequence of practicing SSI provisions of the current seismic design codes on the structural performance is investigated in a probabilistic framework. The soil‐structure system is modeled by the sub‐structure method. The uncertainty in the properties of the soil and the structure is described by random variables that are input to this model. Monte Carlo sampling analysis is employed to compute the probability distribution of the ductility demand of the structure, which is selected as the metrics for the structural performance. In each sample, a randomly generated soil‐structure system is subjected to a randomly selected and scaled ground motion. To comprehensively model the uncertainty in the ground motion, a suite of 3269 records is employed. An extensive parametric study is conducted to cover a wide range of soil‐structure systems. The results reveal the probability that SSI increases the ductility demand of structures designed based on the conventional fixed‐based assumption but built on flexible soil in reality. The results also show it is highly probable that practicing SSI provisions of modern seismic codes increase the ductility demand of the structure. Copyright © 2016 John Wiley & Sons, Ltd.     

IM‐based and EDP‐based decision models for the verification of the seismic collapse safety of buildings

Decision models for the verification of seismic collapse safety of buildings are introduced. The derivations are based on the concept of the acceptable (target) annual probability of collapse, whereas the decision making involves comparisons between seismic demand and capacity, which is familiar to engineering practitioners. Seismic demand, which corresponds to the design seismic action associated with a selected return period, can be expressed either in terms of an intensity measure (IM) or an engineering demand parameter (EDP). Seismic capacity, on the other hand, is defined by dividing the near‐collapse limit‐state IM or EDP by an appropriate risk‐targeted safety factor (γ _im or γ _edp ), which is the only safety factor used in the proposed decision model. Consequently, the seismic performance assessment of a building should be based on the best possible estimate. For a case study, it is shown that if the target collapse risk is set to 10^?4 (0.5% over a period of 50 years), and if the seismic demand corresponds to a return period of 475 years (10% over a period of 50 years), then it can be demonstrated that γ _im is approximately equal to 2.5 for very stiff buildings, whereas for buildings with long periods the value of γ _im can increase up to a value of approximately 5. The model using γ _edp is equal to that using γ _im only if it can be assumed that displacements, with consideration of nonlinear behavior, are equal to displacements from linear elastic analysis.