GNGTS 2011

SESSIONE 2.3

models (all characterized by Vs increasing with depth), we created the 4D function that relates the expected SH-wave amplification factor Fa to (Vs0, f0, Z): we call this the VFZ matrix. We do not really feel the need to fix boundaries between new site classes because this procedure if rigidly instead of statistically interpreted inevitably adds up problems at the class boundaries. However, and just to discuss the benefits of a classification based on the VFZ matrix, we group our ~600 soil modes in terms of the expected amplification at low (< 1 Hz) or high frequency ( 1 Hz) and in terms of low (< 1.5), intermediate (1.5-2) and high (>2) amplification and analyze the performance of this site classification on the response spectra for a input ground motion Ricker wavelet with frequency 1 Hz and 0.5 Hz. For the 1 Hz input motion, we find that the maximum acceleration in the response spectrum is expected on soils with Fa 1.5 and f0 1 Hz, which is intuitive. The minimum acceleration is expected on soils with Fa < 1.5 and f0 < 1 Hz. For the 0.5 Hz input motion, the maximum acceleration is expected on soil classes with Fa 1.5 and f0 < 1 Hz, which is again intuitive. The minimum acceleration is expected on soils with Fa < 1.5 and f0 1 Hz. This shows that if a simplified subsoil classification is needed to infer the Fa and the response spectrum, this should better be based on (Fa, f0) classification scheme rather than on Vs30 classes, which are not unambiguously related neither to Fa, nor to f0. Several methods exist to estimate (Vs0, f0, Z), however the microtremor H/V technique is here preferred because there are no techniques as easy as H/V to get a first order idea of the soil stiffness trends in the subsoil in the whole frequency domain of interest (which is easily > 100 m depth for > 6-8 storey buildings). The microtremor H/V techinique provides a reasonable estimate of f0. Vs0 can be derived from the fit of the H/V curve constrained to the results of a shallow drilling or penetration test (Castellaro and Mulargia, 2009) or from other independent geophysical techniques (usually surface wave based array techniques). Last, Z can be inferred from the H/V peak amplitude A0, through specific modeling. We show that the proposed classification scheme based on (Vs0, f0, Z) can be used also on sites where no specific resonances are measured (due to the absence of sharp impedance contrasts) and on soils presenting several resonances.
References
Castellaro, S., Mulargia, F., 2009a. VS30 estimates using constrained H/V measurements, Bull. Seismol. Soc. Am., 99, 761773.

TIME OCCURRENCE OF EARTHQUAKE INSTABILITIES: A DETERMINISTIC APPROACH

P. Crupi2, A. Bizzarri1, S. de Lorenzo2, M. Loddo2
1 Istituto Nazionale di Geofisica e Vulcanologia, Sezione di Bologna, Italy 2 Dipartimento di Scienze della Terra e Geoambientali, Universit degli Studi di Bari Aldo Moro, Bari, Italy

Understanding the complexity of the fault dynamics and the seismicity recurrence may help improving the earthquake prediction and can give significant advances in seismic hazard assessment. Since the latter half of the last century many studies and laboratory experiments have focused on the understanding of the evolution of frictional resistance during sliding on active faults. Moreover, the increasing knowledge on computational science and numerical modeling has allowed the scientific community to expect more realistic results from the theoretical study of the earthquake events. As pointed out by Bizzarri (2011) numerical models about the mechanisms occurring during faulting represent a valuable tool to explore realistic configurations that are very often far from being fully reproduced in laboratory experiments, due to intrinsic and technical limitations Frictional events may occur through aseismic fault creep, high-velocity or seismic slip and, in some cases, as a combination of both. The complexity of the seismic sliding may stem out from interaction between distinct segments of the same active fault (e.g., asperities), or from strong heterogeneity of shearing resistance. Moreover, according to the concept that earthquakes are frictional instabilities, their time occurrence may show a periodical pattern (the seismic cycle of the so-

352

GNGTS 2011

SESSIONE 2.3

called characteristic earthquake) whose behavior can be referred to the stick-slip dynamic. The complexity and the time recurrence of earthquakes may be examined and consequently modeled with an analog fault system known as the spring-slider model (namely, a damped harmonic oscillator) which simulates the interaction between slip surfaces with the surrounding elastic medium. Despite its obvious limitations, such a model has provided important insights on dynamics of stickslip cycle [Gu et al., 1984], nucleation of earthquakes and triggered earthquake phenomena [e.g. Belardinelli et al., 2003]. The equation of motion governing the spring-slider dynamic system, including the inertial term, in the following: where u is the slip, m is the mass per unit of area, is the frictional resistance, uload is the load point displacement and k is the elastic spring constant. On the basis of several experimental results on rock friction, conducted through the spring-slider system, laboratory-derived constitutive laws have been carried out. In particular in the present study we have considered different analytical expressions of the rate- and state- dependent friction laws formulated by Dieterich [DR henceforth, 1979, equation (2)], and Ruina (RU henceforth, 1983, equation (3)), in which the frictional resistance is expressed as a function of the slip velocity v and a state variable, accounting for the previous slip episodes: (2) (1)

(3) Afterwards, Chester and Higgs (CH henceforth, 1992) figured out that also the temperature variation, produced by frictional heating, can affect the duration of the seismic cycle and the evolution of the frictional strength as well and they consequently incorporated such a thermal effect, improving the previous Ruinas constitutive law: (4) In the above equations is the frictional resistance of a sliding surface which depends on the constant frictional coefficient * considered at a reference slip rate v* and on two governing parameters a and b: the former describes the direct dependence of friction on the slip rate and the latter states the evolution effect that the velocity exerts on the frictional strength as well. Moreover, depends on the characteristic distance L which controls the evolution of the state variables (2) and ((3), (4)), which account for the slip history and for properties of contact sliding surfaces (memory effects). Furthermore v is the slip velocity, neff is the effective normal stress, R is the universal gas constant, T* and T f are respectively the initial temperature and the temperature developed by frictional heating. For a better understanding of the constitutive equations in the mechanics of seismogenic faults the readers can refer to the comprehensive discussion of Bizzarri (2011). The spring slider apparatus combined with the previously described rate- and state- dependent friction laws produces a transient behavior with maximum slip velocities well in excess of the loading velocity. Therefore the present study is aimed to run numerical simulations combining the rateand state- constitutive laws (2), (3) and (4) with the equation of motion of the spring slider apparatus (1), in order to: 1. understand when the system goes through the seismic instability and if it depends on the adopted constitutive law;

353

GNGTS 2011

SESSIONE 2.3

2. show the possible influence that each constitutive laws can exert on the time occurrence of a seismic instability and, hence, on the seismic cycle. The basic idea for modeling the entire process of earthquake cycles consists of a starting tectonic loading, a quasi-static rupture nucleation, the dynamic rupture propagation, the subsequent stress redistribution and the fault strength restoration and so on. Therefore, in order to quantitatively describe the entire earthquake generation cycle it is necessary to have a full description of the motion along the fault, either at low or high speeds. For this purpose according to the simulation strategy employed by Belardinelli et al. (2003), in our simulations the equation of motion is solved either for the quasi-static regime and also for the dynamic one. In particular, as long as the slip rate is lower than critical threshold called critical velocity (v vc) the dynamic equation is solved in the quasi-static approximation (i.e., by neglecting the inertial term in equation (1)), and when v > vc the motion is fully dynamic and the complete equation (1) is solved. Moreover we have also considered a further modification of the models (2), (3) and (4), essentially based on the findings of Weeks (1993), who postulated that the steady state friction becomes independent. of the velocity at high slip rates. Technically, such a behavior consists in assuming what follows: For all three previous constitutive laws (2), (3) and (4) the Weeks approach (5) has been considered stating that the velocity v is frozen (FR, henceforth) at a suitable value of the slider velocity vc as soon as such a critical threshold is exceeded. For completeness we have also considered the not frozen case (NOT FR, henceforth) without any constrains at high slip rates. (5)

Fig. 1 - Slip velocity time history in a linear scale. The light and dark red lines represent the CH simulations, the light and dark green lines represent the RU simulations and the light and dark blue lines show the DR simulations.

Fig. 2 - Slip velocity time history in a logarithmic scale. The light and dark red lines represent the CH simulations, the light and dark green lines represent the RU simulations and the light and dark blue lines show the DR simulations.

354

GNGTS 2011

SESSIONE 2.3

Fig. 3 - Recurrence time (Tcycle) as resulting from the different governing models adopted in the present work. Open symbols refer to the frozen simulation while those closed refer to the not frozen simulation.

In conclusion in this study we have mainly focused on the understanding of the effects that a particular rate- and state-constitutive law can exert on the seismic cycle including its frozen and not frozen constrains comparing the three constitutive laws (DR, RU, CH) through numerical calculations. It was demonstrated here that the seismic cycle duration may depend on: - the specific analytical formulation rate- and state-dependent constitutive law considered to govern the seismogenic fault; - the sliding velocity constrains, whether frozen or not frozen approximation is adopted; - the adopted constitutive parameters (a, b and L). In particular, our results have clearly showed that a spring-slider system governed by the CH constitutive law, compared to the DR and RU ones, needs a lower slip rate peak and a lower time interval to undergo to a seismic instability (Fig. 1, Fig. 2). Indeed, the earthquakes tend to occur earlier in the CH case probably because, such a constitutive model accounts for a further weakening effect due to the temperature evolution (Fig. 1, Fig. 2). Finally, we have also seen that the not frozen simulations promote a time advance of a seismic event and also provides shorter seismic cycle times with respect to the frozen simulations for both the DR and the CH laws (Fig. 3). To finish, we want to highlight that the present results clearly demonstrate that the deterministic prediction of the earthquake occurrence even in the simplest case of a single isolated fault with spatially homogeneous rheology is markedly affected by the non-obvious choice of the constitutive law which describes the time evolution of the frictional resistance. This has a clear impact within the seismic hazard assessment.
Belardinelli M. E., Bizzarri A. and Cocco M.; 2003: Earthquake triggering by static and dynamic stress changes. J. Geophys. Res., 108, No. B3, 2135, doi: 10.1029/2002JB001779. Bizzarri A.; 2010a: Toward the formulation of a realistic fault governing law in dynamic models of earthquake ruptures, in Dynamic Modeling. Edited by Alisson V. Brito, INTECH, Vienna, ISBN 978-953-7619-68-8, 167188, http://www.sciyo.com/articles/show/title/toward-the-formulation-of-a-realistic-fault-governing-law-in-dynamic-models-ofearthquake-ruptures. Bizzarri A.; 2010b: Determination of the temperature field due to frictional heating on a sliding interface. Rapporti Tecnici I.N.G.V., 158, 116. Bizzarri A.; 2010c: On the recurrence of earthquakes: Role of wear in brittle faulting. Geophys. Res. Lett., 37, L20315, doi: 10.1029/2010GL045480. Chester F. M. and Chester J. S.; 1998: Ultracataclasite structure and friction processes of the Punchbowl fault, San Andreas system, California. Tectonophys., 295, 199221. Chester F. M. and Higgs N. G.; 1992: Multimechanism friction constitutive model for ultrafine gouge at hypocentral conditions. J. Geophys. Res., 97, 18591870, doi: 10.1029/91JB02349. Dieterich J. H.; 1979: Modeling of rock friction, 1, Experimental results and constitutive equations. Geophys. Res., 84, 2161 2168. Gu J. C., Rice J., Ruina A. L. and Tse S. T.; 1984: Slip motion and stability of a single degree of freedom elastic system with rate and state dependent friction. J. Mech. Phys. Solids, 32, 167196. Kato N.; 2001: Effect of frictional heating on preseismic sliding: a numerical simulation using a rate, state and temperaturedependent friction law. Geophys. J. Int., 147, 183188 Ruina A. L.; 1983: Slip instability and state variable friction laws. J. Geophys. Res., 88, 10,35910,370. Sibson R. H.; 2003: Thickness of the seismic slip zone. Bull. Seism. Soc. Am., 93, No. 3, 11691178. Weeks J. D.; 1993: Constitutive laws for highvelocity frictional sliding and their influence on stress drop during unstable slip. J. Geophys. Res., 98, 17,63717,648.

References

355