Contents lists available at ScienceDirect Soil Dynamics and Earthquake Engineering journal homepage: www.elsevier.com/locate/soildyn Incremental dynamic based fragility assessment of reinforced concrete structures: Stationary vs. non-stationary artificial ground motions Francesco Basone a , Liborio Cavaleri b , Fabio Di Trapani c, ⁎ , Giuseppe Muscolino d a Università degli Studi di Enna “Kore”, Facoltà di Ingegneria e Architettura, Cittadella Universitaria, 94100 Enna, Italy b Università degli Studi di Palermo, Scuola Politecnica, Dipartimento di Ingegneria Civile, Ambientale, Aerospaziale, dei Materiali DICAM, Viale delle Scienze Ed. 8, 90128 Palermo, Italy c Politecnico di Torino, Dipartimento di Ingegneria Strutturale, Edile e Geotecnica, Corso Duca degli Abruzzi, 24, 10128 Turin, Italy d Università degli Studi di Messina, Dipartimento di Ingegneria Civile, Informatica, Edile, Ambientale e Matematica Applicata, 98122 Messina, Italy ARTICLE INFO Keywords: Artificial accelerograms Non-stationary random processes Nonlinear dynamic analysis Incremental dynamic analysis Fragility curves Reinforced concrete structures ABSTRACT Artificial and natural records are commonly employed by researches and practitioners to perform refined seismic assessments of structures. The techniques for the generation of artificial records and their effectiveness in producing signals which are significantly representative of real earthquakes are still debated as well as results of the consequent seismic assessment to expect from their application. The paper presents an in-depth comparative study highlighting the effect of employing different typologies of artificial ground motion records on seismic assessment results, especially addressing seismic fragility curves. Three sets of 50 stationary, nonstationary evenly modulated and fully nonstationary accelerograms are generated based on design spectrum compatibility criteria. Standard nonlinear time history analyses of 4 reference structural models of reinforced concrete (RC) structures having different degree of complexity are firstly carried out monitoring results in terms significant engineering seismic demand parameters. So far, incremental dynamic analysis (IDA) is used to derive fragility curves. Peak ground acceleration and spectral acceleration are used as possible intensity measures in order to compare results of seismic fragility assessment. The combination of structural irregularity, severe damage and input typology is finally analyzed and discussed in order to assess the degree of dependence of fragility as- sessments on the typology of signal adopted. 1. Introduction Nonlinear dynamic analysis is nowadays increasingly employed as the benchmark for seismic assessment of new and existing construc- tions, in particular when the quantification of the actual seismic de- mand becomes of crucial importance. In this context the selection of ground motions still constitutes a widely debated issue, mainly focusing on advantages and disadvantages associated with the choice of natural or artificial records. On the one hand, artificial records allow efficiently matching of criteria for spectrum compatibility and are easy to be generated. However, standard artificial generation methods provide signals which, differently from natural records, are stationary in am- plitude and frequency. On the other hand, the use of natural ground motions records needs first a large data-set to make proper selections. Moreover scaling of signals is generally necessary to match spectrum compatibility conditions. To date, technical codes (e.g. Eurocode 8 [1], and Italian NTC 2008 [2]) do not provide specifications about the strategies to follow for the generation of artificial records, entrusting the reliability of the selection to the spectrum compatibility criteria. However, they imply the stationarity of artificial accelerograms at least for a given duration, leaving uncertain the possibility to use or not signals generated from nonstationary processes. Several methods have been proposed in the literature in order to generate spectrum compatible artificial accelerograms [e.g. 3,4,5,6] also based on the use of a spectrum compatible power spectral density function (PSD). Response spectra and PSD have in fact a strong re- lationship as highlighted by Vanmarcke and Gasparini [7]. Based on this relationship, a number of procedures (e.g [8–12]) are available in the literature for determining, first, the spectrum compatible PSD and then spectrum compatible signals. The generation of spectrum compa- tible accelerograms filtered by samples of stationary random processes is widely faced in the literature (Shinozuka [13], Barenberg [14], Cacciola et al. [15]). Shinozuka's method provides that samples of spectrum compatible signals can be simulated through the super- position of harmonics with a random phase. In the approach followed by Barenberg [14] a single spectrum compatible accelerogram is http://dx.doi.org/10.1016/j.soildyn.2017.09.019 Received 1 May 2017; Received in revised form 31 August 2017; Accepted 23 September 2017 ⁎ Corresponding author. E-mail address: fabio.ditrapani@polito.it (F. Di Trapani). Soil Dynamics and Earthquake Engineering 103 (2017) 105–117 0267-7261/ © 2017 Elsevier Ltd. All rights reserved. MARK