Академический Документы
Профессиональный Документы
Культура Документы
Summary
This article reviews the ground motion parameters that can be assumed as structural and non-structural damage measures. Measures of seismic damage potential based on ground motion records are rst described, followed by a discussion of the damage measures relating to simple (linear) and more complex (non-linear) structural responses. The second section reviews the measures of damage Prog. Struct. Engng Mater. 2000; 2: 5059 phenomena which govern structural degradation and/or collapse, including experimental results and analytical models. The relationship between earthquake characteristics and type and level of damage on the seismic response of structures is examined, and data from different well-known destructive earthquakes are given.
A key issue in the performance-based design is a reliable assessment of seismic damage potential. If the expected intensity of the earthquake is greatly overestimated, the cost of new construction and seismic rehabilitation of existing structures could be excessive. On the other hand, if the intensity is seriously underestimated, the results may be heavy damage and loss of life. To this end, a reliable definition of seismic intensity has to relate to the effect of damage on structural behaviour in order to assess the potential seismic hazard and to classify the seismic input. Different reports have been published recently on the use of a damage index and damage measures in earthquake engineering. They aim to clarify the different approach methodologies[6**8] and to detail different proposed formulations[911]. This review will focus mainly on the crosscorrelation between ground motion parameters used as damage potential measures and structural and nonstructural damage measures.
Abbreviations EPH = elastoplastic with hardening EPP = elasto-perfectly-plastic ESDOF = elastic single-degree-offreedom EPS = elastoplastic with softening PGA = peak ground acceleration PGD = peak ground displacement PGV = peak ground velocity RC = reinforced concrete
RMSA = root mean square acceleration RMSD = root mean square displacement RMSV = root mean square velocity SDOF = single-degree-of-freedom
Terminology Ae = elastic spectral acceleration ag = ground motion acceleration Ap = plastic spectral acceleration b = low-cycle fatigue parameter = Park & Ang parameter C = amplication factor D = damage index d = displacement dg = ground motion displacement De = elastic spectral displacement Eh = hysteretic energy
EI Fy IA ID IS m n neq o
= = = = = = = = = =
input energy yielding force Arias intensity damage factor Housner spectral intensity average of the plastic cycles, mass ductility number of plastic cycles number of equivalent cycles intensity of zero crossing
PD = R = T = T1,T2 = tB tD tE Ve vg = = = = =
Saragoni index reduction factor elastic period/time periods of Newmark & Hall representation bracketed duration effective duration record duration elastic spectral velocity ground motion velocity
DAMAGE INDICES AND DAMAGE MEASURES Ground motion parameters and damage measures
In this section, damage measures based on data from earthquake records alone, without any structural response data, will be described. The simplest way to characterize the ground motions is in terms of peak or integral parameters. Table 1 reports the values of ground motion parameters for some destructive earthquakes.
Peak parameters
51
shown by the excellent agreement of their correlation studies with the observed damage in the Chilean earthquake of 1985[16]. Cosenza & Manfredi have proposed a damage factor, ID, that is related to the number of plastic cycles, n, and therefore, to the energy content of the earthquake[28]: (4) All the integral measures depend upon the duration of the earthquake which is a measure that cannot be predicted with any certainty. Furthermore, experiences from different earthquakes have confirmed that the duration of ground motion largely influences the level of structural damage. Records with large acceleration and spectral values produce slight damage if the duration is short (eg the Ancona earthquake in 1972), whereas records with low acceleration and long duration can be very destructive (eg the Mexico earthquake in 1985)[17*]. The earthquake duration, tE, is an unpredictable measure, as already mentioned, hence, defining earthquake duration in relation to the main energy content is a problem. In regard to this, Trifunac & Brady[18] have defined the effective duration, tD, as the time elapsed between 5% and 95 % of the RMSA; Kawashima & Aizawa[19] have introduced the bracketed duration, tB, which they define as elapsed time between the first and last acceleration being greater than a percentage of PGA. Trifunac & Novikova[20] have recently proposed a more refined determination of tD as the sum of the record intervals with a total amount of RMSA greater than 90%. Such a definition is more effective for an earthquake characterized by a chain of two or more events, such as the Campano-Lucano earthquake in 1980.
The peak parameters are peak ground acceleration (PGA), velocity (PGV) and displacement (PGD), and peak motion ratios PGV/PGA and PGD/PGV. The PGA is a basic measure of earthquake potential but is not totally reliable. Examination of recorded seismic events has shown that earthquakes with a very large PGA could not produce appreciable structural damage, while earthquakes with a very low PGA could produce an unexpectedly high level of destruction. Instead, the PGV seems to be a more representative measure of earthquake intensity as it is directly connected with energy demand[12]. The peak motion ratio PGV/PGA is indicated by different authors as being a measure of destructiveness, for instance in refs[13,14] the ground motions with larger damage potential show higher PGV/PGA values. Furthermore, this peak ratio can be related to energy demand as discussed below.
Integral parameters
The integral parameters are the root mean square acceleration (RMSA), velocity (RMSV) and displacement (RMSD) defined as: (1) where x(t) is either the ground acceleration ag(t), velocity vg(t) or displacement dg(t) and tE is the total duration of the earthquake. These integral parameters are obviously much more effective for measuring the energy content of a seismic event. Two measures of earthquake destructiveness based on the RMSA are the Arias intensity, IA[15], defined as follows: (2) and the Saragoni Factor, PD[16], given by: P D= I A/ o2 (3)
where the intensity of zero crossing o is the number of zero crossings of the record in the time unit. The Arias intensity can be related to energy content[11], whereas Araya & Saragoni have shown that PD is a more effective measure of earthquake destructiveness (in terms of expected ductility) as
Copyright 2000 John Wiley & Sons, Ltd
52
Table 1 Ground motion parameters for destructive earthquakes
PGA Earthquake Nahanni 1985 Kobe 1995 Chile 1985 Ancona 1972 Friuli 1976 Bucharest 1977 Mexico 1985 CampanoLucano 1980 Record S1 L JMA NS Llolleo N Rocca NS Tolmezzo WE Incerc NS SCT EW Calitri WE (cm/s2) 1080.5 817.8 639.5 538.1 429.3 315.2 192.3 167.9 156.0 PGV (cm/s) 46.2 92.0 41.1 10.9 41.3 32.6 69.0 61.8 20.9 PGD (cm) 10.4 71.8 14.2 5.5 8.2 4.6 18.1 21.9 19.0
PGV/PGA RMSA (s10-2) 4.3 11.2 6.4 2.0 9.6 10.3 35.9 36.8 13.4 (cm/s2) 113.0 88.5 96.3 47.3 112.0 43.0 31.5 29.3 29.6
tD (s) 7.92 8.34 35.68 2.85 10.52 4.93 15.58 38.82 47.17
IA (cm/s) 462.5 838.4 1520.8 67.8 446.2 119.9 71.4 243.8 134.1
o (1/s) 16.93 4.80 9.50 99.03 4.75 5.08 4.36 1.13 4.16
PD (cms) 1.61 36.45 16.85 0.01 19.75 4.66 3.75 189.81 7.77
ID
and of displacement De(T) = CdPGD, in a range of low (with T < T1), medium (T1 < T < T2) and long periods (T > T2), respectively. PGA, PGV and PGD are the peak values of acceleration, velocity and displacement respectively, while Ca, Cv and Cd are the amplification factors. The fundamental periods, T1 and T2, govern the shape of the spectrum. These periods and the amplification factors can be evaluated using the maximum likelihood method from the actual spectrum. For rigid structures (T < 0.03 s), the spectral acceleration is equal to PGA, while for very low periods (0.04 < T < 0.4 T1) it varies with a linear law[22]. An example of the NewmarkHall spectral representation is given in Fig. 1. The NewmarkHall representation is not effective for all types of earthquake (ie for spectra with multiple-peaks or high-amplification periods). However, it allows the spectral shape to be described easily.
The maximum pseudo-acceleration, AeM, is a measure of the maximum strength demand of the earthquake; in fact it is proportional to the maximum seismic force acting on the structure. The maximum pseudo-velocity, VeM, is a measure of the kinetic energy of the earthquake and it is related to the peak energy demand. The Housner spectrum intensity, IS[23], defined as the integral of the pseudo-velocity spectrum in the range 0.12.5 s can be interpreted as an integral measure of the energy demand. Kappos[24] suggested a modification of the period range for better scaling criteria. The values of linear response parameters for some destructive earthquakes are given in Table 2. The maximum displacement, DeM, is a measure of the maximum displacement demand and can be associated with the maximum inter-storey drift[25] which controls structural and non-structural damage
a (m/s2)
53
(cm2/s2) 4611 31675 28600 409 15119 3053 7026 65486 3913
due to lateral displacements in multi-storey buildings. On average, the amplification factors assume values equal to 2.5[22] even though the magnitude, the epicentral distance and the site conditions have a large influence on them. Assuming constant values for the amplification factors, ie not dependent on the earthquake, PGA, PGV and PGD can be assumed to be damage measures instead of the corresponding spectral peak values. As far as the fundamental periods are concerned, high values of T1 represent a great range of structures subjected to the maximum strength demand and, as a result, a large amount of damage; the range T1, T2 is representative of the number of structures subjected to peak energy demand. Ref.[26] proposes that T1 and T2 could be expressed as proportional functions of PGV/PGA and PGD/PGV, respectively: therefore, large values of PGV/PGA correspond to higher values of T1.
analysis. The definition of consistent expressions of R is still an open question and different formulations have been suggested in the literature[27]. The plastic acceleration Ap(T,) represents for a fixed value of the supplied ductility a measure of the strength demand of the ground motion. As will be discussed in the following, this measure is representative of the behaviour of structures that collapse for maximum plastic displacement, independent of the amount of dissipated energy. On the other hand, the cyclic collapse of structures that show a degenerative behaviour and/or cumulative damage is dependent on the amount of hysteretic energy, Eh. The demand of Eh is an effective measure of the damage potential of the earthquake even though a correlation with the ground motion parameters is not simple. The energy dissipation is due to plastic cycles in the structural response. The assessment of the number of plastic cycles, n, and of the average value of their amplitude, m, allows evaluation of the equivalent number of cycles, neq[9,29]. This parameter is a measure of the distribution of amplitude cycles and indicates the number of cycles at the maximum plastic displacement (uy) which the structure has to develop in order to dissipate the total amount of hysteretic energy, Eh. Hence, values of neq close to 1 indicate the presence of one large plastic cycle and are typical of an impulsive earthquake such as the ground motion of Bucharest NS, whereas high values of neq refer to a large number of plastic cycles and are typical of long duration earthquakes as seen in the ground motion records for the Chilean earthquake of 1985. Rodriguez[30] gave a better definion of the parameter neq, referred to as a measure of damage potential. The latter is defined as the number of complete cycles of the equivalent SDOF system with a maximum displacement equal to the peak acceptable roof displacement so as to absorb the total hysteretic energy. The assessment of Eh is performed generally by knowing the input energy EI and the ratio EI/Eh given
Prog. Struct. Engng Mater. 2000; 2: 5059
54
as a function of ductility[31,32]. The input energy can be defined (in the relative formulation) as: (5) where v(t) is the relative velocity and m is the mass of the structure. The relative and the absolute formulations provide very similar results in the period range which is of practical interest[17]. The input energy from a ground motion depends mainly on the elastic period of the structure and on the seismic record, whereas it is slightly influenced by the viscous damping and by the characteristics of the plastic response[31]. For these reasons, the input energy has been assumed as a measure of the energy content of the earthquake[17]. An early relation was suggested by Housner and is valid for the undamped system[33]. It relates the peak value of the input energy, EImax, to the square of the spectral peak velocity. Another basic relationship was proposed by Akiyama[34] which relates the peak input energy to the predominant period of ground motion assumed to be similar to T1. The study of various earthquake records indicates that the peak input energy largely depends on the global work of the earthquake and therefore an integral measure must be included in a reliable formulation for the assessment of EImax[17,26] which relates the effective duration tD to the input energy, and thus tD becomes an effective measure of the intensity of the ground motion. Different formulations for EImax have been given[28,31,35], containing the RMSA and either T1[35] (which can be assumed as proportional to PGV/PGA) or directly to PGV/PGA[28,31]. In this way, the ratio PGV/PGA can be assumed as a measure of the energy content of the earthquake[36]. The values of non-linear response parameters for some destructive earthquakes are given in Table 2, where Ehmaxis evaluated for = 4.
In the non-evolutionary models, the state of the system at time ti+1 only depends on the state at time, ti. Examples of non-evolutionary models are the elasticperfectly-plastic (EPP) and elastoplastic with hardening (EPH) or softening (EPS) models.
Evolutionary models
In the evolutionary models, the state of the system at time ti+1 depends on the state at various earlier times tj. Examples of evolutionary models are the originProg. Struct. Engng Mater. 2000; 2: 5059
55
(c)
(d)
Fig. 2 Hysteretic behaviour: (a) elastoplastic model, (b) evolutionary model (CloughJohnston model), (c) degrading model (stiffness degradation), (d) degrading model (strength degradation)
oriented model, the Clough-Johnston model, the Slip model, the Q-model, etc.
Degrading models
In the degrading models, the state of the system at time ti+1 depends on the states at various earlier times tj and on the values at the time ti of appropriate damage indices. Different degrading models have been proposed for RC or steel structures[3742]. Examples of different hysteretic models are shown in Fig. 2. However, it is interesting to note that in some cases the damage index governing model degradation also governs collapse, but in other cases degradation and collapse depend on different damage indices.
the statistical regression of experimental tests, while in some cases the experimental procedures are not standardized. Different papers have been published recently to detail the different proposed formulations[6,10], mainly regarding RC structures. The most commonly used local indices will be briefly described in the following sections in order to present some case studies. The structural quantities applied to the structures will be indicated in eqns (6) and (7) by the subscript dem, while the supplied quantities will be indicated by the subscript sup. The first applied structural damage parameter is the kinematic or cyclic ductility, , that can be defined in terms of rotation, curvature or displacement. The choice of the kinematic or cyclic ductility as a damage measure is equivalent to assuming that the collapse of the structural model is expected for maximum plastic displacement, independent of the number of plastic cycles and the amount of dissipated energy. Experimental tests have shown that collapse depends on the maximum displacement demand for well-detailed RC elements and structures without bond or shear failure and for steel structures without local buckling phenomena. However, it has been shown that this index can also be used for structures with cumulative deterioration such as in the case of impulse-type or short-duration earthquakes which
Prog. Struct. Engng Mater. 2000; 2: 5059
56
are characterized by one cycle with a large plastic displacement and others with a small amount of plastic work. The plastic dissipated energy, Eh, was first adopted as a damage measure considering cumulative energy absorption[34]. Different researchers have proposed variations of the index, including modification factors, to take into account the specific geometrical and loading conditions of RC[44,45] or steel elements. Experimental evaluation of the capacity of energy absorption is reported in ref.[46]. An improvement to this basic index has been proposed in ref.[47] introducing a more complex rule in the accumulation of energy. The energy-based indices are very attractive for their simplicity in the evaluation of dissipated energy in a structure as observed above, but the experimental assessment of the supplied energy dissipation capacity is very difficult. Another approach is based on the accumulation of damage due to cyclic loading that is usually modelled by introducing the low-cycle fatigue law. A general deformation cumulative index can be defined by introducing the well-known Coffin & Mason[48,49] and Miner[50,51] laws: (6) The value assumed by this damage index is defined in terms of constant b, which depends on the structural material and typology, and on the number of different plastic displacements, independent of the exact order of the displacements. Typical values of b, obtained by experimental data for steel structures[52] and RC structures[7,53], are 1.61.8; in a damage analysis, sometimes a conservative value of 1.5 is assumed[54]. Experimental values for different elements are recorded in ref.[55]. For b = 1 the index gives the same weight to each plastic displacement, independent of their individual values; hence, DF coincides with the energy-based index DE for the EPP model. For high values of b (ie > 5) the low-cycle fatigue index provides results similar to the ductility index[8]. The potential of the CoffinMason law to analyse low-cycle fatigue in steel components is shown in ref.[56], and in a RC section in ref.[53]. A generalized version of the cumulative index has been proposed in ref.[41] for RC structures considering separately positive and negative cycles and including weight factors in Miners rule. Different formulations of the cumulative index are proposed in refs[42,53] for concrete structures and in refs[57,58] for steel structures. The damage index based on low-cycle fatigue has the advantage of taking into account the distribution of the amplitude of plastic cycles and of having clear experimental tests to assess numerical parameters. Other indices are based on a combination of
Copyright 2000 John Wiley & Sons, Ltd
57
G a low-cycle fatigue damage index with b > 5, in practice, provides the same result as the ductility method; with higher values of this parameter it seems logical to use the simple ductility index method. G the Park & Ang and low-cycle fatigue damage indices, with suggested values of the parameters ( = 0.15 and b = 1.8), provide results which are, in practice, coincident. The analysis of seismic behaviour of the CloughJohnston, origin-oriented and Slip models, (nonevolutionary, degrading), compared with the response of the EPP model[37], shows (see Fig. 4): G the behaviour of structures governed by the ductility damage function seems to be only slightly influenced by the hysteretic models, G the behaviour of structures controlled by the energy damage function is more sensitive to modelling. The analysis of the evolutionarydegrading model proposed in ref.[37] shows that the effect of strength degradation and of re-loading stiffness degradation is
very important, and structures which show this kind of damage have to be designed using accelerations which are greater than those corresponding to nondegrading structures, while the effect of unloading stiffness degradation is less important. Therefore, it is worth pointing out that the structures characterized by strength degradation must be designed with particular attention; the structures with large slips due to bond and buckling of bars (RC structures) or with early local buckling (steel or composite structures) fall within this class. On the other hand, structures characterized only by unloading stiffness degradation can be simulated using non-degrading models like the CloughJohnston model, with small errors in the analysis.
Conclusions
Evaluation of damage measures and damage index is a very complex topic due to the complexity of the seismic input and the strong influence of structural response. In this review an attempt has been made at a systematic classification of recent papers on this subject.
(b)
(a)
Fig. 4 Inuence of hysteretic behaviour on seismic response: (a) functional cyclic ductility damage, (b) functional energy damage
A collection of papers that represents the state of art on the nonlinear behaviour of structures under seismic loadings.
[2] Abrams DP & Calvi GM (eds). Proceedings USItalian Workshop on Seismic Evaluation and Retrot. Technical Report NCEER-97-0003. Buffalo, New York: National Center for Earthquake Engineering Research. 1997. [3] Bertero VV. State of art report on design criteria. In: Proceedings 11th World Conference on Earthquake Engineering, Acapulco, Mexico, 2328 June 1996. ** [4] Fajfar P & Krawinkler H (eds). Seismic design methodologies for the next generation of codes. Rotterdam: Balkema. 1997.
A collection of papers that represents the state of art in new trends in design methodologies related to the performance-based approach.
* [5] SEAOC Vision 2000 Committee. Performance based seismic design engineering. Sacramento, USA: Structural Engineers Association of California (SEAOC) Report. 1995.
This article gives a well-stated presentation of the different approach methodologies in the denition and evaluation of a damage index for RC structures.
[7] Powell GH & Allahabadi R. Seismic damage prediction by deterministic methods: concepts and procedures. Earthquake Engineering and Structural Dynamics 1988: 16: 719734. [8] Cosenza E, Manfredi G & Ramasco R. The use of damage functionals in earthquake-resistant design: a comparison among different procedures. Structural Dynamics and Earthquake Engineering 1993: 22: 855868. [9] McCabe SL & Hall WJ. Assessment of seismic structural damage. Journal of Structural Engineering (ASCE) 1989: 115(9): 21662183. [10] Williams MS & Sexsmith RG. Seismic damage indices for concrete structures: a state of the art review. Earthquake Spectra 1995: 11(2): 319349. [11] Fardis M. Damage measures and failure criteria for reinforced concrete members. In: Proceedings 10th European Conference on Earthquake Engineering, Vienna, 1995. Rotterdam: Balkema. 1995. 13771382. [12] Housner GW & Jennings PC. Earthquake design criteria. EERI Monograph Series. Berkeley, CA: Earthquake Engineering Research Institute. 1982. [13] Zhu TJ, Tso W K & Heidebrecht AC. Effect of peak ground a/v ratio on structural damage. Journal of Structural Engineering (ASCE) 1988: 114: 10191037. [14] Meskouris K, Kratzig WB & Hanskotter U. Seismic motion damage potential for R/C wall-stiffened buildings. In: Fajfar P & Krawinkler H (eds) Nonlinear seismic analysis and design of reinforced concrete buildings. Oxford: Elsevier Applied Science. 1992. 125136. [15] Arias A. A measure of earthquake intensity. In: Seismic design of nuclear power plants. Cambridge, MA: MIT Press. 1970. 438468. [16] Saragoni GR. Response spectra and earthquake destructiveness. In: Proceedings 4th US National Conference on Earthquake Engineering, Palm Springs, USA, 2024 May 1990. Berkeley, CA: Earthquake Engineering Research Institute (EERI). 1990. 3543. * [17] Uang CM & Bertero VV. Evaluation of seismic energy in structures. Earthquake Engineering and Structural Dynamics 1990: 19: 7790.
59
[58] Calado L & Castiglioni CA. Steel beam-to-column connection under low cycle fatigue: experimental and numerical research. In: Proceedings 11th World Conference on Earthquake Engineering, Acapulco, Mexico, 2328 June 1996. [59] Park YJ. Seismic damage analysis and damage-limiting design of R/C structures. PhD Thesis, Department of Civil Engineering, University of Illinois, Urbana, IL. 1984. [60] Park YJ & Ang AH-S. Mechanistic seismic damage model for reinforced concrete. Journal of Structural Engineering (ASCE) 1985: 111(4): 722739. [61] Park YJ, Ang AH-S & Wen YK. Damage-limiting aseismic design of buildings. Earthquake Spectra 1987: 3(1): 126. [62] Chai YH, Romnstadt KM & Bird SM. Energy based linear damage model for high intensity seismic loading. Journal of Structural Engineering (ASCE) 1995: 121(5): 857864. [63] Banon H & Veneziano D. Seismic safety of reinforced concrete members and structures. Earthquake Engineering and Structural Dynamics 1982: 10: 179193.
Edoardo Cosenza PhD Professor, Dipartimento di Analisi e Progettazione Strutturale, Universit` a di Napoli Federico II, Via Claudio 21, 80125 Napoli, Italy E-mail: cosenza@unina.it Gaetano Manfredi PhD Associate Professor, Dipartimento di Analisi e Progettazione Strutturale, Universit` a di Napoli Federico II, Via Claudio 21, 80125 Napoli, Italy