Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

The emergence of crack-like behavior of frictional rupture: Edge singularity and energy balance

The emergence of crack-like behavior of frictional rupture: Edge singularity and energy balance The emergence of crack-like behavior of frictional rupture: Edge singularity and energy balance 1 2 1 Fabian Barras , Michael Aldam , Thibault Roch , E m 3;4 2  1y A. Brener , Eran Bouchbinder , and Jean-Fran cois Molinari Civil Engineering Institute, Materials Science and Engineering Institute, Ecole Polytechnique F ed erale de Lausanne, Station 18, CH-1015 Lausanne, Switzerland Chemical and Biological Physics Department, Weizmann Institute of Science, Rehovot 7610001, Israel Peter Grunb  erg Institut, Forschungszentrum Julich,  D-52425 Julich,  Germany Institute for Energy and Climate Research, Forschungszentrum Julich,  D-52425 Julich,  Germany The failure of frictional interfaces | the process of frictional rupture | is widely assumed to feature crack-like properties, with far-reaching implications for various disciplines, ranging from engineering tribology to earthquake physics. An important condition for the emergence of a crack- like behavior is the existence of stress drops in frictional rupture, whose basic physical origin has been recently elucidated. Here we show that for generic and realistic frictional constitutive relations, and once the necessary conditions for the emergence of an e ective crack-like behavior are met, frictional rupture dynamics are approximately described by a crack-like, fracture mechanics energy balance equation. This is achieved by independently calculating the intensity of the crack-like singularity along with its associated elastic energy ux into the rupture edge region, and the frictional dissipation in the edge region. We further show that while the fracture mechanics energy balance equation provides an approximate, yet quantitative, description of frictional rupture dynamics, interesting deviations from the ordinary crack-like framework | associated with non-edge-localized dissipation | exist. Together with the recent results about the emergence of stress drops in frictional rupture, this work o ers a comprehensive and basic understanding of why, how and to what extent frictional rupture might be viewed as an ordinary fracture process. Various implications are discussed. I. BACKGROUND AND MOTIVATION  =  , the di erence between the applied driving d res stress  and the residual stress  , in frictional rup- d res ture. In a very recent paper [1] we showed that, contrary Rapid slip along interfaces separating bodies in fric- to widely adopted assumptions, the residual stress res tional contact is mediated by the spatiotemporal dynam- is not a characteristic property of frictional interfaces. ics of frictional rupture [41, 43], which is a fundamental Rather, for rapid rupture  is shown to crucially de- res process of prime importance for a broad range of physi- pend on elastodynamic bulk e ects | in particular wave cal systems. For example, it is responsible for squealing radiation from the frictional interface to the bulks sur- in car brake pads [36], for bowing on a violin string [16], rounding it and long-range elastodynamic bulk interac- and for earthquakes along geological faults [8, 28, 35], tions | and that the existence of a nite stress drop  , to name just a few well-known examples. A very power- is a nite time e ect, limited by the wave travel time in ful conceptual and quantitative framework to understand nite systems. Speci cally, it has been shown that frictional dynamics in a wide variety of physical contexts is the analogy between frictional rupture and ordinary ( ) ' v ( ) ; (1) fracture/cracks. d d res 2c This framework is extensively used to interpret and quantify geophysical observations [2, 12], as well as a where  is the shear modulus of the bulks surrounding the broad spectrum of laboratory phenomena [7, 26, 27, 34, frictional interface, c is the corresponding shear wave- 40, 42, 44, 45]. For example, a recent series of careful lab- speed and v is the theoretically predicted residual slip res oratory experiments [7, 44, 45] demonstrated that when velocity behind the propagating rupture edge. v ( ) is res the analogy between frictional rupture and ordinary frac- determined through the approximate equation  (v ) + ss res ture holds, the dynamic propagation of laboratory earth- v ' , once long-range elastodynamic contributions res 2c quakes and their arrest can be quantitatively understood are omitted [1], where  (v) is the steady-state friction ss to an unprecedented degree [24]. Yet, the fundamental curve as a function of slip velocity v. physical origin and range of validity of the analogy be- The theoretical prediction in Eq. (1) has been sup- tween frictional rupture and ordinary fracture are not yet ported by existing experimental results for rapid fric- fully understood. tional rupture [1], for times shorter than the waves re- An important condition for the analogy to hold is ection time from outer boundaries, and by computer the emergence of a nite and well-de ned stress drop simulations in in nite systems. An example taken from one of these computer simulations is presented in Fig. 1a (cf. Fig. 3 in Barras et al. [1]), where two rapid rup- eran.bouchbinder@weizmann.ac.il ture fronts propagating in opposite directions are ob- jean-francois.molinari@ep .ch served, leaving behind them a well-de ned stress drop arXiv:1907.04376v2 [cond-mat.soft] 20 Nov 2019 2 that quantitatively agrees with the theoretical pre- rupture propagation is shown to be non-edge-localized, dictions (see Barras et al. [1] for details). The most out- i.e. to be spatially extended, and as such demonstrates standing theoretical question that remains open in the interesting deviations from the ordinary crack-like frame- context of the analogy between frictional rupture and or- work. Finally, the signi cance and implications of our dinary cracks, once the necessary conditions associated ndings for various phenomena are brie y discussed in with the emergence of a nite stress drop  are met, Sect. IV. Together with the recent results about the is to what extent the analogy actually holds, both in emergence of stress drops in frictional rupture [1], this qualitative and in quantitative terms. This question is work o ers a comprehensive and basic understanding of systematically addressed in this paper. why, how and to what extent frictional rupture might be viewed as an ordinary fracture process. The existence of a nite stress drop  does not im- mediately guarantee that the analogy between frictional rupture and ordinary fracture holds because proper scale separation should also be satis ed. That is, the resid- (a) ual stress  behind the propagating rupture should be res reached on a scale (typically termed the cohesive zone) 0.38 much smaller than the rupture size L (cf. Fig. 1a). If such scale separation is valid, we expect all crack-like properties to emerge in frictional rupture. In particu- 0.36 lar, we expect the frictional stress and slip velocity elds near the rupture edge to feature the famous square root singularity of conventional fracture mechanics [8]. More- 0.34 over, under these conditions, we expect the singularity- -4 -2 0 2 4 associated energy ux into the edge region to balance the edge-localized energy dissipation in excess of the power (b) 0.45 invested against the residual stress  . This energy bal- res ance relation amounts to an e ective equation of motion 0.4 for rupture propagation [8]. 0.35 In this paper we show that for generic and realistic frictional constitutive relations, and once the conditions 0.3 -7 -5 -3 -1 0 for the emergence of an e ective crack-like behavior are 10 10 10 10 10 met, frictional rupture dynamics are approximately | yet quantitatively { described by a crack-like, fracture mechanics energy balance equation [8]. This is achieved FIG. 1. (a) A snapshot of the frictional stress  (x) (normal- ized by the normal stress ) during rupture propagation that in a few steps. In Sect. II we show that if one assumes emerges in dynamic simulations with the steady-state friction the existence of the conventional square root singularity law shown in panel (b) and  = 0:375 (see text and Bar- of ordinary fracture mechanics and the associated near- ras et al. [1] for additional details). The snapshot reveals edge energy balance in frictional rupture, the latter fol- two rapid rupture fronts (the rupture length L is marked) lows a generic rupture length-velocity relation based on propagating at an instantaneous speed c ' 0:84c in oppo- r s the knowledge of the stress drop  alone. In Sect. III, site directions into regions characterized by the applied stress we quantitatively and systematically test these assump- and leaving behind them a well-de ned residual stress tions separately. We rst show that the conventional <  . Consequently, a well-de ned and nite stress drop res d square root singularity of standard fracture mechanics  emerges, as marked. Note that the y-axis is truncated provides a good quantitative description of the near rup- at = = 0:4 for visual clarity and that x is normalized by a generalized Grith-like length L , de ned in Eq. (6) (with ture edge stress and slip velocity elds simultaneously. a unity prefactor). (b) The steady-state friction stress  (v), ss We then propose a physically-motivated procedure to in- normalized by a constant normal stress , vs. the slip rate dependently extract an e ective fracture energy from the v (solid brown line). The curve has a generic N -shape [13], dissipative interfacial dynamics and show that it is bal- with a maximum at an extremely low v and a minimum at anced by the singularity-associated energy ux into the an intermediate v. The horizontal line represents the driving edge region to a good approximation. stress  , which intersects the N -shaped steady-state friction These results indicate that the scale separation men- curve at three points; the leftmost and rightmost ones are tioned above is approximately satis ed for frictional rup- stable xed points, while the intermediate one is an unstable one. The e ective steady-state friction curve (dash-dotted ture and that indeed the e ective fracture energy corre- orange line) is obtained by adding v (with  = 9GPa and sponds to edge-localized dissipation. However, the pro- 2c c = 2739m/s) to the solid brown line, see Barras et al. [1] for posed procedure to extract the relevant edge-localized s more details. The stress drop  of Eq. (1), which equals the dissipation allows us to show, also in Sect. III, that there one shown in panel (a), is marked by the black double-arrow. exists additional energy dissipation in excess of the power invested against the residual stress  . This contribu- res tion to the energy dissipation associated with frictional 3 II. CRACK-LIKE SCALING AND THE the energy release rate [23], even though it is not a rate), DEPENDENCE OF THE LENGTH-VELOCITY taking the form [23] RELATION ON THE STRESS DROP [K (L; c )] G(L; c )  A(c =c ) ; (4) r r s As explained above, and with the results of Barras et al. [1] in mind, we aim at carefully exploring the impli- where A(c =c ) is a known universal and dimensionless r s cations of stress drops | once they exist | for frictional function that depends on the fracture symmetry mode dynamics. The expected implications, to be detailed be- (here mode-II or mode-III). Finally, by invoking energy low, directly follow from the analogy to ordinary frac- balance in the edge region, standard fracture mechanics ture mechanics and consequently from its standard pre- predicts that [8] dictions [8, 43]. The challenge is to test whether these predictions are satis ed as emergent properties of the un- G(L; c ) = G (c ) ; (5) r c r derlying physics without assuming them a priori. Some of these predictions have been previously studied in the where G (c ) is the e ective fracture energy (of dimen- c r literature [10, 15{17, 19, 33, 46], but to the best of our sions of energy per unit area) associated with the transi- knowledge these studies have not yet led to a compre- tion from the v 0 state ahead of the edge to the v > 0 hensive picture of the analogy between frictional rupture state behind it, which possibly depends on the rupture and ordinary fracture. speed c . It is crucial to understand that unlike ordinary The existence of a stress drop behind the two edges tensile (mode-I symmetry) fracture, where G (c ) is the c r of propagating frictional rupture, cf. Fig. 1a, suggests only dissipation in the problem, in the friction problem that the load bearing capacity of the interface in this frictional dissipation exists everywhere along the sliding region is reduced,  < , and consequently that parts res d interface and not just in the transition region near the of the interface ahead of the edges should compensate for rupture edge. The way energy dissipation is partitioned this reduction, i.e. carry stress that is larger than  . In in the friction problem will be discussed below. the framework of the classical theory of fracture, the so- The above discussion raises several basic questions; called Linear Elastic Fracture Mechanics (LEFM), this most notably, does the square root singularity of Eqs. (2)- stress ampli cation ahead of the rupture edges follows a (3) generically exist in frictional rupture once  ex- universal singularity as the rupture edge is approached [8] ists? Can the e ective fracture energy G (c ) be mean- c r ingfully separated from the entire dissipation associated K (L; c ) (x)  p ; K (L; c )   LK(c =c ) ; (2) with frictional motion? And if so, can the energy bal- r r s jx x j ance of Eq. (5) be veri ed by independently calculating both G and G (the latter using Eq. (4))? While various where K quanti es the intensity of the singularity (hence aspects of these questions have certainly been addressed it is termed the stress intensity factor [23]), x is the lo- in the literature [10, 15{17, 19, 33, 46], we believe that cation of each of the rupture edges, L is the instanta- systematically addressing all of them in a single system is neous distance between the two edges (i.e. the rupture still missing. Before performing such a systematic anal- length/size, cf. Fig. 1a) and K(c =c ) is a dimension- r s ysis, we address rst a rather strong implication of the less function of the instantaneous propagation speed c of relations discussed above. each edge. We note that here and below numerical pre- Combining Eqs. (2)-(5), one obtains the following factors are omitted as we are interested in crack-like scal- stress drop dependent length-velocity relation ing relations in this section. In addition, the slip velocity is predicted to follow the very same singular behavior c =c = F [L=L ( )] with L ( )  ; (6) r s G G ( ) c K (L; c ) r r v(x)  p ; (3) jx x j which is valid under the assumption that G is indepen- dent of c . Here L ( ) is a generalized Grith-like r G just behind the edges (note the absolute value). As length [7, 8] and F () is a monotonically increasing func- expected, the intensity of the ampli cation/singularity tion that we do not specify. K (L; c ) in Eq. (2) increases with increasing  and the To test this prediction, we employed the generic rate- rupture length L (L is the size of the region in which and-state friction constitutive framework, presented in the interfacial load bearing capacity is reduced, hence a detail in Barras et al. [1]. Within this framework, the larger compensation/ampli cation exists). The relations interfacial constitutive law at any position x along the in Eqs. (2)-(3) are valid independently of the symme- interface and at any time t is described by the following try mode of rupture, and in particular in the context of local relation frictional rupture, they are valid for both in-plane shear (mode-II) and anti-plane shear (mode-III) symmetries. =  sgn(v) f (jvj; ) ; (7) Standard fracture mechanics predicts that the square root singularity in Eqs. (2)-(3) is accompanied by a nite which must be supplemented with a dynamical equation ux of energy G into the rupture edge region (known as for the evolution of . Extensive evidence indicates that 4 1 f (jvj;  = D=v) =  (v)=, under steady-state sliding ss ss conditions and a controlled normal stress , has been measured over a broad range of slip rates v for many materials [11]. 0.9 Together with general theoretical considerations [13], it is now established that the steady-state frictional stress (v) is generically N -shaped, as shown in Fig. 1b (solid ss brown line). Finally, the e ective friction curve obtained by adding the radiation damping term v, which has 0.8 2c been shown to play an important role in the emergence of stress drops in frictional rupture [1], is also presented in Fig. 1b (dash-dotted orange line). We would like to 0.5 0.6 0.7 0.8 stress that, as shown in Barras et al. [1], pure velocity- weakening friction laws also e ectively feature N -shaped behavior due to the radiation damping term (and hence also feature a nite stress drop). Consequently, the re- sults to be presented below equally apply to velocity- weakening friction laws. 0.9 Coupling this constitutive framework to spectral boundary integral method [5, 22, 30] calculations in in- nite systems under mode-III deformation conditions, gave rise to frictional rupture such as the one shown 0.8 in Fig. 1a. In this approach, the displacement eld ^ ^ ^ ^ u(x; y; t) =u (x; y; t)z (the unit vectors satisfy z? x; y) is computed at the interface y! 0 self-consistently with 5 10 15 20 the far- eld stress  and the friction law of Eq. (7), see Barras et al. [1] for additional details. Based on such numerical computations, we plot in Fig. 2a the nor- FIG. 2. (a) The frictional rupture velocity c , normalized by malized frictional rupture velocity c =c vs. the frictional r r s c , as a function of the frictional rupture length L (normal- s rupture length L for various driving stress levels  (de- ized by the system size W = 320m used in these calculations) tailed in the legend of Fig. 2b). The di erent c (L) curves for di erent driving stress levels  = (as detailed in the leg- span a rather broad range. Equation (6) predicts that end of panel (b)), using the N -shaped friction law of Fig. 1b. these curves can be collapsed onto a master curve if L is Frictional rupture is nucleated as described in Barras et al. rescaled by L ( ), where  ( ) is given in Eq. (1) (see G d [1]. (b) The prediction of Eq. (6) is tested by plotting c =c r s also Fig. 3c in Barras et al. [1]) and the e ective fracture vs. L=L ( ), where  varies with  according to Eq. (1) G d energy G is assumed to be independent of c . To follow c r (see also Fig. 3c in Barras et al. [1]). L ( ), as de ned in this rescaling procedure, L ( ) of Eq. (6) is evaluated Eq. (6), is evaluated with  = 9GPa, G = 0:65J/m and a with a unity prefactor,  = 9GPa and G = 0:65J/m . unity prefactor. The length-velocity curves of panel (a) all c The way to extract the value of the e ective fracture en- collapse on a master envelope curve as predicted by Eq. (6), see additional discussion in the text. ergy G is discussed in Sect. III below. The outcome of the rescaling procedure is presented in Fig. 2b. physically represents the age/maturity of the contact (hence it is related to the real contact area) [9, 11, 20, It is observed that the di erent c (L) curves, which ex- 29, 31, 32, 38], and that its evolution takes the form hibited a rather large spread in Fig. 2a, collapse on the envelope of a single master curve upon rescaling L by jvj L ( ). Note that deviations from the master curve are = g ; (8) observed at early times (small L values in each curve); this is expected as the crack-like behavior cannot be valid with g(1) = 0 and where  is of time dimension. The in the nucleation stage, but rather only when L is suf- characteristic slip displacement D controls the transition ciently large and frictional rupture is suciently well- from a stick state v 0, with a characteristic structural developed. The collapse in Fig. 2b provides indirect, yet state  =  , to a steadily slipping/sliding state v > 0, strong, support to the applicability of the crack-like rela- with  =D=v. The precise functional form of g() (with tions in Eqs. (2)-(5) to frictional rupture. These relations ss g(1) = 0) plays no role in what follows. The function will be directly tested next. 5 III. THE EMERGENCE OF STRESS for the di erent locations x's at small , but exhibits lo- SINGULARITY AND LOCAL ENERGY cation dependence at signi cantly larger , where it levels BALANCE o to di erent limiting values that become closer to one another as x increases. These observations can be under- stood as follows; the frictional stress  (x; t) presented in One of the major implications of the existence of a - Fig. 3a exhibits two distinct behaviors behind the prop- nite stress drop  is the emergence of stress singularity agating rupture edge (here the propagation is from right near the frictional rupture edge, as explained above and to left). First, it features a strong decay well within the as formulated in Eqs. (2)-(3). In order to directly test this edge region. Second, as denoted by the arrow, there ex- prediction, we present in Fig. 3a the (properly normal- ists a transition to a slow decay towards  on a signi - ized) spatial pro les of  (x; t) and v(x; t) near a rupture res cantly larger lengthscale, extending far beyond the edge edge at time t. We then t the two elds together to region (the full spatial extent of this decay is not shown). Eqs. (2)-(3), demanding the same stress intensity factor This slow spatial decay stems from the rate and state de- K and the same e ective tip location x (the details of pendence of the friction law, which implies that all of the tting procedure are extensively discussed in the SM the interfacial elds in the problem  (x; t); v(x; t); (x; t) [1]). slowly approach their respective asymptotic steady-state The resulting ts are superimposed on the elds  (x; t) values  ; v ; D=v . Finally, as rupture propagation res res res and v(x; t) in Fig. 3a. The square root singular behav- in the presence of a nite stress drop is intrinsically out ior faithfully describes the two elds near the front edge, of steady state, i.e. rupture accelerates towards c as supporting the prediction that such a singular behavior shown in Fig. 2, we expect some position dependence emerges in the presence of a nite stress drop  . Note of E (; x). This dependence should become weaker as BD that the spatial range in which the elds are described by the limiting velocity c ! c is approached, as is indeed the square root singular behavior is larger for the slip ve- r s observed in Fig. 3b. locity v(x; t) than for the frictional stress  (x; t). The rea- The physical picture emerging from the above discus- son is that  (x; t) features a signi cantly narrower range sion suggests that the location independent part of the of values between its peak value and the applied stress breakdown energy E (; x), which is associated with BD (in the large jxj limit) compared to the corresponding excess dissipation near the rupture edge, should be iden- range for v(x; t), and thus the latter can accommodate a ti ed as the e ective fracture energy G appearing in singular behavior, which is by construction an interme- Eq. (5). This idea is pictorially demonstrated by the diate asymptotic behavior, over a larger spatial range. horizontal black line in Fig. 3b, which identi es G with The results of Fig. 3a demonstrate that a rather well- the point in which the various E (; x) curves start to BD de ned stress intensity factor K (L; c ) is associated with split/deviate one from another (from which a value of frictional rupture in the presence of a nite stress drop G  0:65J/m can be inferred). To make the identi ca- , from which the energy release rate G(L; c ) can be tion of G more quantitative and to allow a direct test readily extracted using Eq. (4) [1]. Next, in order to test of Eq. (5), we invoke the observation that the combina- the validity of Eq. (5), we need to independently calculate tion v=D strongly overshoots unity in the edge region the e ective fracture energy G associated with frictional (v=D> 1 implies < 0, which is associated with contact rupture propagation. To this aim, we de ne the energy area reduction), then slightly undershoots it and nally per unit area that is dissipated at a given interfacial loca- approaches unity from below far from the edge [1]. We tion x during the transition from a non-slipping/sticking note that the position of the rst crossing v=D = 1 ap- state to a steadily sliding state characterized by the resid- proximately corresponds to the position marked by small ual stress  [10] res arrow in Fig. 3a. Consequently, the edge-localized dis- sipation G can be estimated as the excess dissipation 0 0 E (; x) =  ( )  d : (9) BD res associated with the spatial region for which v=D > 1, quanti ed by the following spatial integral Here the slip history at a location x is given by the slip G (c )   (x; t)  v(x; t) dx : (10) displacement (x; t)  u (x; y = 0 ; t)u (x; y = 0 ; t), c r res z z c (t) r v=D>1 where (x; t) =v(x; t), and the subscript 'BD' stands for 'breakdown'. The breakdown energy quanti es the excess We note that this estimate of G appears to be consis- dissipation on top of the frictional dissipation associated tent with an analytic approximation available in the lit- with sliding against the residual stress  . Note that we erature [15{17], which may shed light on the dependence res cannot a priori identify the breakdown energy de ned in of G on interfacial parameters (see SM [1] for details). Eq. (9) with the e ective fracture energy G , as will be We are now in a position to directly test Eq. (5), where discussed next. the energy release rate G is calculated using the stress In Fig. 3b we plot the breakdown energy E (; x) at intensity factor extracted as shown in Fig. 3a and G BD c 4 di erent interfacial locations x =` , i = 14, ordered by through Eq. (10). In the inset of Fig. 3b, we plot the their proximity to the nucleation site (the center of the ratio G=G as a function of the rupture length L. It is domain). It is observed that E (; x) perfectly overlaps observed that G=G is close to unity throughout the rup- BD c 6 0.25 1.2 -1 0.2 0.8 -2 0.15 -4 -3 -2 10 10 10 1.25 0.6 0.1 0.4 0.75 0.05 0.2 0.5 0.2 0.4 0.6 0.8 0 0 -0.004 -0.002 0 0.002 0.004 0 0.1 0.2 0.3 0.4 0.5 FIG. 3. (a) The normalized spatial pro les of  (x; t) and v(x; t) near a rupture edge propagating from right to left with a velocity c ' 0:94c at time t. x is shifted by x , which corresponds to the location of e ective rupture edge (cf. Eqs. (2)-(3)). r s r 2 2 Both elds are normalized/shifted by quantities de ned in the text, except for  1 c =c . The dashed lines are the results r s 1=2 of tting the solid lines to Eqs. (2)-(3), with K = 64kPam , see SM [1] for additional details. The tilted arrow is discussed in the text. (inset) The same as the main panel, but on a double logarithmic scale with the x-axis being jx x j. Note that since the dashed lines in the main panel are symmetric with respect to x , using jxx j implies the existence of a single dashed r r line in the inset. The inset highlights both the quality of the t and the di erent spatial ranges used for each eld, see SM [1] for additional details. (b) The breakdown energy E (; x), de ned in Eq. (9), vs. slip  for 4 interfacial locations x =` , with BD i ` =W = 0:15, ` =W = 0:20, ` =W = 0:25 and ` =W = 0:30. ` are measured from the nucleation site (the center of the system) 1 2 3 4 i and the system size is W = 80m. The horizontal black line marks the splitting of the di erent curves, which is identi ed with G  0:65J/m . (inset) G=G vs. L=W , where L is the rupture length. G is calculated using K (L), cf. panel (a) and the SM c c [1], through Eq. (4) and G is calculated through Eq. (10). The generic properties of the results presented in this gure are independent of the details of the friction law (not shown). ture propagation history, lending strong support to the IV. SUMMARY AND CONCLUDING REMARKS ideas developed above. In particular, it shows that the rupture edge energy balance in Eq. (5) provides quanti- tative approximations for frictional rupture dynamics. In this paper we set out to further explore the analogy between frictional rupture and ordinary fracture. The At the same time, our results also clearly demon- starting point for this investigation is our own very recent strate that E (; x) can be quite signi cantly larger BD work that elucidated the physical origin of stress drops than G and position dependent, implying that non- in frictional rupture [1], which constitute a necessary edge-localized dissipation in excess of the power invested condition for the analogy. Our major goal was to under- against the residual stress  is a generic property of res stand to what extent the analogy holds, both in quali- frictional interfaces featuring rate and state dependent tative and in quantitative terms, for interfaces described friction. A similar physical situation has been discussed by generic and realistic frictional constitutive relations, in Brener and Marchenko [15]. That is, while a physi- once stress drops do exist. cally sensible extraction of the edge-localized excess dis- sipation G allows to obtain reasonably well quantitative We showed that for rate and state constitutive rela- approximations for frictional rupture dynamics based on tions, frictional rupture dynamics are approximately | the analogy to ordinary fracture, our results clearly in- yet quantitatively | described by an ordinary fracture dicate that this analogy is incomplete and that interest- energy balance equation, when the conditions for the ing deviations exist. These deviations are intimately re- emergence of a nite stress drop  are satis ed. To es- lated to the spatially extended (non-edge-localized) rate tablish the quantitative status of this fracture mechanics and state dependence of frictional interfaces, an intrin- energy balance equation, we proposed a physical criterion sic frictional property that is entirely absent in ordinary for extracting the rupture edge-localized dissipation di- fracture, and are manifested in non-edge-localized excess rectly from the frictional dynamics, allowing to de ne an dissipation. The latter may have important implications e ective fracture energy G for frictional problems. Sur- for the energy budget associated with frictional dynam- prisingly, we discovered that G does not account for all ics, and might be relevant to geophysical observations of the energy dissipation E in excess of the energy dis- BD and their interpretations [17, 19, 33, 46]. sipated against the residual stress  (cf. Eq. (9)). These res 7 ndings imply that the analogy between frictional rup- et al. [17] and Tinti et al. [46], which concluded based on ture and ordinary fracture is not complete, as manifested seismic data that the breakdown energy can be larger by the existence of a non-edge-localized contribution to than the fracture energy for large earthquake ruptures. E . These results o er insight into open questions concerning BD earthquake energy budget [2, 17, 19, 33, 46] and deserve The di erence between E and G is intimately re- additional investigation. BD c lated to the generic rate and state dependence of friction, More generally, we expect our results to provide which is responsible for the two-step nature of the stress a conceptual and quantitative framework to address relaxation/weakening process associated with frictional various fundamental and applied problems in relation rupture propagation; rst, there exists a rather sharp to the rupture dynamics of frictional interfaces, with stress drop that takes place over a relatively small slip, implications for both laboratory and geophysical-scale bringing the stress close to, but not identically to, the phenomena. For example, our results and theoretical residual stress  . Second, there exists a slower, longer- framework are expected to apply also to slip pulses. res term process that brings the stress to the residual stress Indeed, recent preliminary results, see Fig. S6 in Brener over signi cantly larger slip. The latter stress relax- et al. [2], support this expectation. res ation/weakening process, which some authors attribute to melting or thermal pressurization [37, 47] not taken Acknowledgements E. B. and J.-F.M. acknowledge sup- into account in the present work, is responsible for the port from the Rothschild Caesarea Foundation. E. B. ac- di erence between E and G . This physical picture knowledges support from the Israel Science Foundation BD c is reminiscent of the model proposed in Kanamori and (Grant No. 295/16). J.-F.M., F. B. and T. R. acknowl- Heaton [25], and further discussed in Abercrombie and edge support from the Swiss National Science Founda- Rice [2], in trying to resolve some puzzling observations tion (Grant No. 162569). This research is made possible in relation to the energy budget of earthquake rupture. in part by the historic generosity of the Harold Perlman Moreover, this physical picture is consistent with Chester Family. [1] See Supplemental Material for additional information, . progressive evolutionary changes, and di erent dy- [2] Abercrombie, R.E., Rice, J.R.. Can observa- namic regimes. Rev Geophys 2008;46(4):RG4006. tions of earthquake scaling constrain slip weaken- URL: http://doi.wiley.com/10.1029/2008RG000260. ing? Geophys J Int 2005;162(2):406{424. URL: doi:doi:10.1029/2008RG000260. https://academic.oup.com/gji/article-lookup/ [9] Bhattacharya, P., Rubin, A.M.. Frictional re- doi/10.1111/j.1365-246X.2005.02579.x. doi:doi: sponse to velocity steps and 1-D fault nucleation un- 10.1111/j.1365-246X.2005.02579.x. der a state evolution law with stressing-rate dependence. [7] Andrews, D.J.. Rupture propagation with nite stress J Geophys Res Solid Earth 2014;119(3):2272{2304. in antiplane strain. J Geophys Res 1976;81(20):3575{ URL: http://doi.wiley.com/10.1002/2013JB010671. 3582. URL: http://doi.wiley.com/10.1029/ doi:doi:10.1002/2013JB010671. JB081i020p03575. doi:doi:10.1029/JB081i020p03575. [10] Bizzarri, A.. On the relations between fracture [13] Bar-Sinai, Y., Spatschek, R., Brener, E.A., Bouch- energy and physical observables in dynamic earth- binder, E.. On the velocity-strengthening behav- quake models. J Geophys Res 2010;115(B10):B10307. ior of dry friction. J Geophys Res Solid Earth URL: http://doi.wiley.com/10.1029/2009JB007027. 2014;119(3):1738{1748. URL: http://doi.wiley.com/ doi:doi:10.1029/2009JB007027. 10.1002/2013JB010586. doi:doi:10.1002/2013JB010586. [16] Bizzarri, A., Cocco, M.. Slip-weakening behavior [1] Barras, F., Aldam, M., Roch, T., Brener, E.A., Bouch- during the propagation of dynamic ruptures obeying binder, E., Molinari, J.F.. The emergence of crack- rate- and state-dependent friction laws. J Geophys Res like behavior of frictional rupture: The origin of stress 2003;108(B8):2373. URL: http://doi.wiley.com/10. drops. To appear in Physical Review X 2019. URL: http: 1029/2002JB002198. doi:doi:10.1029/2002JB002198. //arxiv.org/abs/1906.11533. arXiv:1906.11533. [12] Bizzarri, A., Liu, C.. Near- eld radiated wave eld [11] Baumberger, T., Caroli, C.. Solid friction from may help to understand the style of the supershear stickslip down to pinning and aging. Adv Phys transition of dynamic ruptures. Phys Earth Planet 2006;55(3-4):279{348. URL: http://www.tandfonline. Inter 2016;261:133{140. URL: https://linkinghub. com/doi/abs/10.1080/00018730600732186. doi:doi: elsevier.com/retrieve/pii/S0031920116300838. 10.1080/00018730600732186. doi:doi:10.1016/j.pepi.2016.05.013. [7] Bayart, E., Svetlizky, I., Fineberg, J.. Frac- [5] Breitenfeld, M.S., Geubelle, P.H.. Numeri- ture mechanics determine the lengths of interface rup- cal analysis of dynamic debonding under 2D in- tures that mediate frictional motion. Nat Phys plane and 3D loading. Int J Fract 1998;93(1/4):13{ 2015;12(2):166{170. URL: https://www.nature.com/ 38. URL: http://link.springer.com/10.1023/A: articles/nphys3539. doi:doi:10.1038/nphys3539. 1007535703095. doi:doi:10.1023/A:1007535703095. [8] Ben-Zion, Y.. Collective behavior of earth- [2] Brener, E.A., Aldam, M., Barras, F., Molinari, J.F., quakes and faults: Continuum-discrete transitions, Bouchbinder, E.. Unstable Slip Pulses and Earthquake 8 Nucleation as a Nonequilibrium First-Order Phase 2010;115(12):1{25. doi:doi:10.1029/2009JB006833. Transition. Phys Rev Lett 2018;121(23):234302. URL: [28] Marone, C.. Laboratoty-derived friction https://link.aps.org/doi/10.1103/PhysRevLett. laws and their application to seismic faulting. 121.234302. doi:doi:10.1103/PhysRevLett.121.234302. Annu Rev Earth Planet Sci 1998;26(1):643{ [15] Brener, E.A., Marchenko, V.I.. Frictional shear 696. URL: http://www.annualreviews.org/doi/ cracks. J Exp Theor Phys Lett 2002;76(4):211{214. abs/10.1146/annurev.earth.26.1.643. doi:doi: URL: http://link.springer.com/10.1134/1.1517386. 10.1146/annurev.earth.26.1.643. doi:doi:10.1134/1.1517386. [29] Marone, C.. The e ect of loading rate on [16] Casado, S.. Studying friction while playing the vi- static friction and the rate of fault healing during olin: exploring the stickslip phenomenon. Beilstein the earthquake cycle. Nature 1998;391(6662):69{72. J Nanotechnol 2017;8:159{166. URL: http://www. URL: http://www.nature.com/nature/journal/v391/ beilstein-journals.org/bjnano/content/8/1/16. n6662/abs/391069a0.html. doi:doi:10.1038/34157. doi:doi:10.3762/bjnano.8.16. [30] Morrissey, J.W., Geubelle, P.H.. A numerical scheme [17] Chester, J.S., Chester, F.M., Kronenberg, A.K.. for mode III dynamic fracture problems. Int J Numer Fracture surface energy of the Punchbowl fault, San Methods Eng 1997;40(7):1181{1196. URL: https: Andreas system. Nature 2005;437(7055):133{136. //doi.org/10.1002/(SICI)1097-0207(19970415)40: URL: http://www.nature.com/articles/nature03942. 7{%}3C1181::AID-NME108{%}3E3.0.CO;2-X. doi:doi: doi:doi:10.1038/nature03942. 10.1002/(SICI)1097-0207(19970415)40:7<1181::AID- [15] Cocco, M., Bizzarri, A.. On the slip- NME108>3.0.CO;2-X. weakening behavior of rate- and state dependent con- [31] Nagata, K., Nakatani, M., Yoshida, S.. A re- stitutive laws. Geophys Res Lett 2002;29(11):1516. vised rate- and state-dependent friction law obtained URL: http://doi.wiley.com/10.1029/2001GL013999. by constraining constitutive and evolution laws sepa- doi:doi:10.1029/2001GL013999. rately with laboratory data. J Geophys Res Solid Earth [19] Das, S.. Dynamic fracture mechanics in the study of the 2012;117(B2):B02314. URL: http://doi.wiley.com/ earthquake rupturing process: theory and observation. 10.1029/2011JB008818. doi:doi:10.1029/2011JB008818. J Mech Phys Solids 2003;51(11-12):1939{1955. URL: [32] Nakatani, M.. Conceptual and physical clari - https://linkinghub.elsevier.com/retrieve/pii/ cation of rate and state friction: Frictional slid- S0022509603001443. doi:doi:10.1016/j.jmps.2003.09.025. ing as a thermally activated rheology. J Geo- [20] Dieterich, J.H.. Applications of rate-and state- phys Res Solid Earth 2001;106(B7):13347{13380. dependent friction to models of fault slip and earthquake URL: http://doi.wiley.com/10.1029/2000JB900453. occurrence. Treatise Geophys 2007;4:107{129. doi:doi: doi:doi:10.1029/2000JB900453. 10.1073/pnas.93.9.3787. [33] Nielsen, S.B., Spagnuolo, E., Smith, S.A.F., Vio- [8] Freund, L.B.. Dynamic Fracture Mechanics. Cambridge: lay, M., Di Toro, G., Bistacchi, A.. Scaling in nat- Cambridge university press, 1998. ural and laboratory earthquakes. Geophys Res Lett [22] Geubelle, P., Rice, J.R.. A spectral method 2016;43(4):1504{1510. URL: http://doi.wiley.com/ for three-dimensional elastodynamic fracture prob- 10.1002/2015GL067490. doi:doi:10.1002/2015GL067490. lems. J Mech Phys Solids 1995;43(11):1791{1824. [34] Noda, H., Lapusta, N., Kanamori, H.. Com- URL: http://linkinghub.elsevier.com/retrieve/ parison of average stress drop measures for ruptures pii/002250969500043I. doi:doi:10.1016/0022- with heterogeneous stress change and implications for 5096(95)00043-I. earthquake physics. Geophys J Int 2013;193(3):1691{ [23] Irwin, G.R.. Analysis of stresses and strains near the end 1712. URL: http://gji.oxfordjournals.org/cgi/doi/ of a crack traversing a plate. J Appl Mech 1957;24:361{ 10.1093/gji/ggt074. doi:doi:10.1093/gji/ggt074. 364. [35] Ohnaka, M.. The physics of rock failure and earthquakes. [24] Kammer, D.S., Radiguet, M., Ampuero, J.P., Moli- Cambridge University Press, 2013. nari, J.F.. Linear Elastic Fracture Mechanics Predicts [36] Rhee, S., Jacko, M., Tsang, P.. The role of friction the Propagation Distance of Frictional Slip. Tribol Lett lm in friction, wear and noise of automotive brakes. 2015;57(3):23. URL: http://link.springer.com/10. Wear 1991;146(1):89{97. URL: http://linkinghub. 1007/s11249-014-0451-8. doi:doi:10.1007/s11249-014- elsevier.com/retrieve/pii/004316489190226K. 0451-8. arXiv:1408.4413. doi:doi:10.1016/0043-1648(91)90226-K. [25] Kanamori, H., Heaton, T.H.. Microscopic and [37] Rice, J.R.. Heating and weakening of faults dur- macroscopic physics of earthquakes. In: Geocomplex- ing earthquake slip. J Geophys Res Solid Earth ity Phys. Earthquakes. American Geophysical Union 2006;111(B5):B05311. URL: http://doi.wiley.com/ (AGU); 2000. p. 147{163. URL: http://www.agu. 10.1029/2005JB004006. doi:doi:10.1029/2005JB004006. org/books/gm/v120/GM120p0147/GM120p0147.shtml. [38] Rice, J.R., Ruina, A.L.. Stability of Steady Frictional doi:doi:10.1029/GM120p0147. Slipping. J Appl Mech 1983;50(2):343{349. URL: [26] Lu, X., Lapusta, N., Rosakis, A.J.. Pulse- http://appliedmechanics.asmedigitalcollection. like and crack-like dynamic shear ruptures on fric- asme.org/article.aspx?articleid=1406945. doi:doi: tional interfaces: experimental evidence, numerical 10.1115/1.3167042. modeling, and implications. Int J Fract 2010;163(1- [17] Rubin, A.M., Ampuero, J.P.. Earthquake nucleation on 2):27{39. URL: http://link.springer.com/10.1007/ (aging) rate and state faults. J Geophys Res Solid Earth s10704-010-9479-4. doi:doi:10.1007/s10704-010-9479-4. 2005;110(B11):B11312. URL: http://doi.wiley.com/ [27] Lu, X., Rosakis, A.J., Lapusta, N.. Rupture 10.1029/2005JB003686. doi:doi:10.1029/2005JB003686. modes in laboratory earthquakes: E ect of fault prestress [40] Rubino, V., Rosakis, A.J., Lapusta, N.. Understanding and nucleation conditions. J Geophys Res Solid Earth dynamic friction through spontaneously evolving labo- 9 ratory earthquakes. Nat Commun 2017;8(7260):15991. URL: http://www.nature.com/doifinder/10.1038/ ncomms15991. doi:doi:10.1038/ncomms15991. [41] Scholz, C.H.. The mechanics of earthquakes and faulting. Cambridge university press, 2002. [42] Svetlizky, I., Bayart, E., Cohen, G., Fineberg, J.. Frictional Resistance within the Wake of Frictional Rup- ture Fronts. Phys Rev Lett 2017;118(23):234301. URL: http://link.aps.org/doi/10.1103/PhysRevLett.118. 234301. doi:doi:10.1103/PhysRevLett.118.234301. [43] Svetlizky, I., Bayart, E., Fineberg, J.. Brittle Frac- ture Theory Describes the Onset of Frictional Motion. Annu Rev Condens Matter Phys 2019;10(1):031218{ 013327. URL: https://www.annualreviews.org/doi/ 10.1146/annurev-conmatphys-031218-013327. doi:doi: 10.1146/annurev-conmatphys-031218-013327. [44] Svetlizky, I., Fineberg, J.. Classical shear cracks drive the onset of dry frictional mo- tion. Nature 2014;509(7499):205{208. URL: https://www.nature.com/articles/nature13202. doi:doi:10.1038/nature13202. [45] Svetlizky, I., Pino Mun ~oz, D., Radiguet, M., Kam- mer, D.S., Molinari, J.F., Fineberg, J.. Prop- erties of the shear stress peak radiated ahead of rapidly accelerating rupture fronts that mediate fric- tional slip. Proc Natl Acad Sci 2016;113(3):542{ 547. URL: http://www.pnas.org/lookup/doi/10.1073/ pnas.1517545113. doi:doi:10.1073/pnas.1517545113. [46] Tinti, E., Spudich, P., Cocco, M.. Earth- quake fracture energy inferred from kinematic rup- ture models on extended faults. J Geophys Res 2005;110(B12):B12303. URL: http://doi.wiley.com/ 10.1029/2005JB003644. doi:doi:10.1029/2005JB003644. [47] Viesca, R.C., Garagash, D.I.. Ubiquitous weaken- ing of faults due to thermal pressurization. Nat Geosci 2015;8(11):875{879. URL: https://www.nature.com/ articles/ngeo2554. doi:doi:10.1038/ngeo2554. S1 Supplemental Material for: \The emergence of crack-like behavior of frictional rupture: Edge singularity and energy balance" The goal of this document is to provide additional tech- nical details regarding the extraction of the near-edge singular elds (Fig. 3a in the manuscript) and the ef- 0.5 fective fracture energy G from the interfacial dynamics (Fig. 3b in the manuscript), both discussed in Sect. III of the manuscript. This is achieved in two steps; rst, in 2 4 6 8 10 Sect. S-1, some relevant concepts and methodology are being discussed and tested using a conventional cohesive zone model of ordinary fracture. Then, in Sect. S-2, these concepts and tools are generalized for frictional rupture along interfaces described by generic friction constitutive relations, and additional details about their application -4 -3 -2 -1 0 1 2 3 4 in Sect. III of the manuscript are brie y provided. The numerical tools and the generic interfacial constitutive re- lation (including the material parameters) are presented FIG. S1. Space-time diagram of the dynamic mode-III rup- in [S1, S2]. ture event described in the text. The yellow region corre- sponds to the broken interface left behind the propagating rupture edges, the narrow red region corresponds to the co- S-1. EDGE SINGULARITY AND ENERGY hesive zone and the black region corresponds to the intact interface. The blue line marks the instant at which the snap- BALANCE IN A CONVENTIONAL COHESIVE ZONE MODEL OF ORDINARY FRACTURE shots of the stress and slip velocity elds in Fig. S2a are taken. (inset) The time evolution of the rupture speed c as function of its size L. Our goal here is to rst develop the procedure for extracting the near-edge singular elds in a simpler case, where there is no residual stress (i.e. ordinary the center of an interface at rest under a uniform shear fracture), where the Linear Elastic Fracture Mechanics stress  , where 0 <  <  , by progressively increasing d d c (LEFM) singularity is regularized on a small lengthscale an originally in nitesimal seed crack toward a critical (i.e. proper scale separation is realized) and the frac- size L = L . The latter, known as the Grith criti- ture energy G is prescribed. This is achieved by the cal length [S7, S8], is given by (see also Eq. (6) in the well-known framework of cohesive zone crack models, at- manuscript) tributed to Dugdale [S3] and Barenblatt [S4], which be- came very popular in the numerical modeling of dynamic 4 G L = ; (S3) fracture (see, for example, [S5, S6]). Within this frame- G work, we employ a linear slip-weakening cohesive law in str which the strength of the interface  linearly reduces for mode-III cracks. In Fig. S1, we present the resulting to zero over a characteristic slip displacement dynamics that feature a crack that progressively accel- erates toward c , the maximal admissible rupture speed str s (x; t) =  f1 (x; t)= g ; (S1) c c for mode-III symmetry. The instantaneous rate of dissipated energy associated where  is the failure strength (determining the rupture with the propagation of one rupture edge (recall that peak stress), (x; t) is the slip displacement, and fg = there are two of these) can be obtained as [S6] if  > 0 and 0 otherwise ( is a dummy variable used to de ne the function fg in Eq. (S1)). The linear slip- Z 1 weakening law of Eq. (S1) corresponds to a prescribed E (t) =  (x; t) v(x; t) dx ; (S4) diss value of the fracture energy 1 where W is the system size. The integral attains a nite G = d =   : (S2) c c c contribution only inside the well-de ned cohesive zone near the propagating rupture edge, where both  (x; t) The spectral boundary integral method under mode- and v(x; t) are non-zero. The cohesive zone (also termed III symmetry (where the basic object is the out-of-plane fracture process zone in ordinary fracture), which cor- displacement eld at the interface, u (x; y = 0; t), see responds to the region where the stress  (x; t) drops manuscript and references therein for details) can be cou- from the peak stress (failure strength)  to 0, is marked pled to Eq. (S1) (i.e. the latter replaces the friction law by the red-shaded region in Fig. S2a. A snapshot of used in the manuscript) to generate propagating rup- the stress  (x; t) and slip velocity v(x; t) distributions ture fronts. In this context, rupture is nucleated at near the propagating rupture edge are also presented in S2 Fig. S2a (and see also Fig. S1). The fracture energy, de- in Fig. S2a. That is, we t the normalized and shifted ned in Eq. (S2), is the energy dissipated per unit crack near-edge stress and slip velocity elds to the singular extension dL form in Eqs. (S6)-(S7), with x and K as the two free r III parameters. To make the procedure well de ned, we also d dE E (t) dL diss diss need to specify the spatial range over which the ts are G (t) = E (t) = = ; (S5) c diss dL dt dt c (t) r performed. In determining the spatial range of the t of the two elds, several physical considerations are invoked; which is constant for the slip-weakening model used here rst, it is clear that the ts cannot include the regions (see Fig. S2b). where the elds (cf. the examples in Fig. S2a) attain their Standard fracture theory predicts that close to the peak values as these are associated with the regulariza- propagating rupture edges, we have the famous square tion of the singular behavior (the cohesive zone). Second, root singular elds [S8] the tting ranges cannot extend too far away from the edge region as the elds there include also non-singular III (r =x x;  = 0; c )  ' p (S6) r r res contributions. Finally, as the overall variability of the 2(x x) stress eld is smaller compared to that of the slip veloc- ity eld, we expect the singular region to be narrower for and the former. We employ a nonlinear least-squares regres- (c ) K s r III sion tting procedure [S9] to determine the best estimates v(r =xx ;  =; c ) ' p ; (S7) r r 2c 2(x x ) for x and K , and selected the tting ranges to be as r r III large as possible within the constraints imposed by the where (r; ) is a polar coordinate system moving with physical considerations just stated. 2 2 the rupture edge, (c ) = 1 c =c , x is the e ective s r r r s edge location and K is the mode-III stress intensity III factor. We subtracted the residual stress  from the res The resulting ts, i.e. the right-hand-sides of Eqs. (S6)- frictional stress eld such that the shifted stress eld (S7), are superimposed on the normalized slip velocity v vanishes behind the rupture edge and normalized the and frictional stress  elds in Fig. S2a (dashed lines). slip velocity eld such that the left-hand-sides of both To highlight the spatial tting ranges used, we replot the Eqs. (S6)-(S7) attain comparable values; note that for results in Fig. S2a on a double logarithmic scale against the slip-weakening model used here we have  = 0, res jx x j=W in the inset (note that due to the symmetry and it makes no di erence, but in general one may have r of the singular form on the right-hand-sides of Eqs. (S6)- > 0 (also in the framework of slip-weakening models), res (S7), we have now a single t that describes the two elds see Sect. S-2. In addition, we used v = 2u _ since v is the over di erent spatial ranges). The inset shows that the slip velocity, not the particle (mass) velocity u _ . Finally, spatial tting ranges for the two elds are di erent, that as is evident from the right-hand-sides of both Eqs. (S6)- the range for the slip velocity eld is wider than the one (S7), the normalized slip velocity v and frictional stress for the frictional stress eld and that the peak regions are elds are symmetric functions relative to x (i.e. it is the properly excluded. Finally, we veri ed that the values very same function of jxx j), though the spatial ranges of x and K are robust against changes in the spatial r III in which the singular form is valid di er for the two elds. tting ranges within the stated constraints. This issue will be discussed below, where we explain how the two free parameters in Eqs. (S6)-(S7) | x and K r III | are determined. We stress that the proper normaliza- tion and shift used in Eqs. (S6)-(S7) allow us to consider The extracted value of K has been used to calculate III the stress and slip velocity elds on equal footing. the energy release rate G according to Eq. (S8). Then we The square root singularity is associated with a nite applied the tting procedure to the whole rupture prop- energy ux into the edge region, the so-called energy agation history and the a priori known value of G in release rate G, which for mode-III symmetry takes the Eq. (S2) has been used to plot in Fig. S2b G=G as a form [S8] function of L=L , where L is the rupture length. The results strongly support the expected relation G=G = 1 1 K III G(t) = : (S8) and hence also validate our tting procedure. Note that some deviation from G=G = 1 is observed, re ecting Our goal now is to extract the stress intensity factor from some uncertainly in the singular behavior, even in simple the singular elds of Eqs. (S6)-(S7), to use Eq. (S8) to slip-weakening models. Finally, for completeness, we also calculate G and to check whether the near-edge energy plot in Fig. S2b E (t)=c (t) of Eq. (S5), normalized by diss r balance G =G is satis ed. As all of the assumptions of G , which indeed equals unity throughout the rupture c c conventional fracture theory are satis ed by the model, propagation process, as expected. The same tting pro- the energy balance equation should be satis ed. cedure is applied in the manuscript to the frictional rup- We start by estimating the stress intensity factor from ture dynamics of interfaces described by rate-and-state the near-edge stress and slip velocity distributions shown friction, as discussed next. S3 1.4 0.8 -1 0.6 -4 -3 -2 10 10 10 0.4 0.6 0.2 0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 2 4 6 8 10 12 FIG. S2. (a) A snapshot of the normalized stress and slip velocity elds (see legend and the left-hand-sides of Eqs. (S6)-(S7)) near the edge of a rupture propagating at a speed c to the left (the snapshot corresponds to the blue horizontal line in Fig. S1, where rupture propagation in the simple slip-weakening cohesive zone model is presented). Note that  is used to nondimensionalize the elds and that  = 0 in this case. The black dashed lines correspond to ts to Eqs. (S6)-(S7), see text res for additional details. (inset) The same as the main panel, but on a double logarithmic scale and the x-axis is jx x j=W , see text for additional details. (b) G and E =c , both normalized by G , are plotted as a function of the normalized rupture size diss r c L=L (see legend in order to distinguish the di erent curves). These quantities are discussed in detail in the text. S-2. APPLICATION TO THE FRICTIONAL th length of Eq. (S3) takes the form RUPTURE DYNAMICS OF INTERFACES 4 G DESCRIBED BY RATE-AND-STATE FRICTION L = ; (S9) ( ) which is identical to the corresponding expression in A procedure similar to the one described in the pre- Eq. (6) in the manuscript, up to the dimensionless and vious section is applied in the manuscript to the fric- order unity pre-factor 4=. tional rupture dynamics of interfaces described by rate- As discussed in the manuscript, the generalized and-state friction. However, the di erences between the Grith-like length in Eq. (S9) and in Eq. (6) in the simple slip-weakening cohesive zone model discussed in manuscript highlights another di erence between simple the previous section and the more realistic rate-and-state slip-weakening cohesive zone models and rate-and-state friction models discussed in the manuscript, which are friction models related to G . While in slip-weakening intimately related to the central question addressed in cohesive zone models G is an a priori prescribed quan- the manuscript, call for some modi cations that will be tity, in rate-and-state friction models the existence and discussed here. First, frictional rupture features a nite identi cation of a well-de ned G from the interfacial dy- residual stress  > 0 under some conditions (extensively namics is not obvious. That is, one should understand res discussed in [S1]). That is, the strength of the inter- whether and how an e ective fracture energy G can be face does not drop to zero behind the rupture front as in properly de ned, and what the associated lengthscale is. the simple slip-weakening cohesive zone model (note that A procedure to de ne and extract G is discussed and in general slip-weakening cohesive zone models can de - employed in the manuscript. Here we supplement it with nitely feature a constant residual stress  ), but rather additional rationalization and details. res attains a nite value (on what lengthscale this value is The basic idea is related to the observation that the attained is yet another central question addressed in the frictional stress  (x; t) follows two distinct relaxation manuscript). The linearity of the elastodynamic eld regimes in the wake of rupture fronts, as demonstrated equations [S10] implies that the driving stress  in the in Fig. 3a in the manuscript. It rst undergoes a rather ordinary fracture case should be simply replaced by the strong initial drop that is followed by a slow decay to- stress drop  =  > 0 in the frictional case. This wards  . Such behavior is inherent to the rate-and- d res res implies that  should be subtracted from the stress state dependence of the frictional strength [S11]. The res eld  (x; t) before tting it to the square root singular initial strong drop is associated with a rather localized contribution in Eq. (S6) (cf. Fig. 3a in the manuscript). region near the rupture edge (see arrow in Fig. 3a in the Moreover, this implies that a generalization of the Grif- manuscript) and the slow decay towards  is charac- res S4 0.25 1.2 ing to v(x; t)(x; t)=D > 1, as formulated in Eq. (10) in the manuscript. 1.15 0.2 The latter criterion is demonstrated in Fig. S3, where 1.1 the frictional stress  (x; t) of Fig. 3a in the manuscript 0.15 -0.01 0 0.01 is superimposed on v(x; t)(x; t)=D, to exactly corre- 1.05 spond to the change in the relaxation behavior of  (x; t) 0.1 towards  that was discussed above. This criterion res is also in line with recent physics-based interpretations 0.05 of rate-and-state friction formulations [S12{S14]. Fi- 0.95 nally, for completeness, we present in Fig. S4 a snap- 0 0.9 shot of the spatial distribution of the real contact area -0.05 -0.04 -0.03 -0.02 -0.01 0 0.01 0.02 0.03 0.04 0.05 A (x; t) 1 + b log[1 + (x; t)= ] [S11]. 1.1 1.2 FIG. S3. A snapshot of the properly normalized (see legend) 1.6 stress eld  (x; t) (left y-axis) and v(x; t)(x; t)=D (right y- 1.15 1.4 1.08 axis) corresponding to the solution presented in Fig. 3a in the 1.2 manuscript, where the y-axis is truncated to allow the prop- 1.1 erties of the elds near the rupture edge to be visible. (inset) 1.06 -0.2 -0.1 0 0.1 v(x; t)(x; t)=D near the rupture edge without truncating the 1.05 y-axis. 1.04 1.02 0.95 terized by a much larger lengthscale. We consequently proposed that the former should be associated with the 0.9 e ective fracture energy G . -0.05 -0.025 0 0.025 0.05 In order to formalize this idea and to make the extrac- tion of G quantitative, we focus on the dimensionless FIG. S4. A snapshot of the real contact area A (x; t) 1 + combination v(x; t)(x; t)=D, which is shown in Fig. S3 b log[1 + (x; t)= ] (blue line, left y-axis) corresponding to and which according to Eq. (8) in the manuscript con- v(x; t)(x; t)=D of Fig. S3, which is reproduced here (orange trols the evolution of the structural state of the inter- line, right y-axis). The real contact area also exhibits slow face (x; t). The latter is known to determine the real relaxation to its asymptotic value behind the rupture edge. contact area A (x; t) 1 + b log[1 + (x; t)= ] of the in- (inset) A full scale plot of A (x; t) 1 + b log[1 + (x; t)= ] terface [S12] (for the de nition of the parameters b and near the rupture edge, directly demonstrating that the latter , and their values used here, see [S1, S2]). Hence, it is is associated with a reduction of the real contact area. directly related to the rupture process, involving a tran- sition from an initial value of A ahead of the rupture We note that the estimation of G through the dissi- front to a signi cantly lower value behind it (see the in- pation corresponding to the criterion v(x; t)(x; t)=D> 1 set of Fig. S4). This transition corresponds to a transi- appears to be consistent with available analytic approx- tion between v=D = 1 ahead of the rupture front, with imations for the e ective fracture energy [S15{S17]. In a very small v and hence a large , and v=D = 1 be- particular, the expression hind it, with a large v and hence a much smaller . In between, v=D is expected to attain signi cantly larger D @f (jvj; ) G = [log(v =v )] (S10) c c bg values. This physical picture is demonstrated in the inset 2 @ log () of Fig. S3, which corresponds to the rupture front shown in Fig. 3a in the manuscript. has been proposed in [S17]. Here @f (jvj; )=@ log () is The two-step nature of the approach of v=D to its the aging coecient (f (jvj; ) is the friction law intro- steady-state is revealed in the main panel of Fig. S3, duced in Eq. (7) in the manuscript), v corresponds to bg which presents a zoomed in version of the inset. The the steady-state velocity in the stick state (prior to the gure reveals that after the huge peak in v=D, which oc- arrival of the rupture front) and v is the slip velocity curs on a small lengthscale near the rupture edge, v=D far behind the rupture front. We estimate v as the bg undershoots unity and then approaches unity slowly from leftmost intersection point in Fig. 1b in the manuscript, below, on a signi cantly larger lengthscale. We conse- i.e. v  10 m/s, and v as the rightmost intersec- bg c quently attribute the small lengthscale weakening pro- tion point with the e ective steady-state friction curve, cess to the near-edge dissipation G , i.e. to the e ec- i.e. v  10 m/s. Using the parameters used in this c c 6 6 tive fracture energy, where the additional dissipation as- work (see [S2]), i.e. D = 0:5 10 m,  = 10 Pa and sociated with the larger lengthscale is discussed in the @f (jvj; )=@ log () = 0:021 (the latter equals bf in the manuscript. In quantitative terms, this picture implies notation of [S2]), and plugging everything in Eq. (S10), that G is estimated through the dissipation correspond- we obtain G  0:7J/m . The latter is in reasonably good c c S5 agreement with G of Fig. 3b in the manuscript. In or- tribution of near-edge elds and to test the energy bal- der to further substantiate this agreement, future work ance relation G =G presented in Sect. S-1 is applied in should extend the comparison by systematically varying the manuscript to rate-and-state frictional interfaces. In the parameters involved. this case,  is replaced by the stress drop  and G d c is estimated from the interfacial dynamics according to To conclude, the procedure to extract the singular con- Eq. (10) in the manuscript, as explained in detail here. [S1] F. Barras, M. Aldam, T. Roch, E. A. Brener, E. Bouch- [S10] A. C. Palmer and J. R. Rice, The Growth of Slip Sur- binder, and J.-F. Molinari, The emergence of crack- faces in the Progressive Failure of Over-Consolidated like behavior of frictional rupture: The origin of stress Clay, Proc. R. Soc. A Math. Phys. Eng. Sci. 332, 527 drops, To appear in Physical Review X (2019). (1973). [S2] E. A. Brener, M. Aldam, F. Barras, J.-F. Moli- [S11] T. Baumberger and C. Caroli, Solid friction from stick- nari, and E. Bouchbinder, Unstable Slip Pulses and slip down to pinning and aging, Adv. Phys. 55, 279 Earthquake Nucleation as a Nonequilibrium First-Order (2006). Phase Transition, Phys. Rev. Lett. 121, 234302 (2018). [S12] T. Baumberger and P. Berthoud, Physical analysis of [S3] D. Dugdale, Yielding of steel sheets containing slits, J. the state- and rate-dependent friction law. II. Dynamic Mech. Phys. Solids 8, 100 (1960). friction, Phys. Rev. B 60, 3928 (1999). [S4] G. Barenblatt, The Mathematical Theory of Equilib- [S13] Y. Bar-Sinai, R. Spatschek, E. A. Brener, and E. Bouch- rium Cracks in Brittle Fracture, Adv. Appl. Mech. 7, binder, On the velocity-strengthening behavior of dry 55 (1962). friction, J. Geophys. Res. Solid Earth 119, 1738 (2014). [S5] M. S. Breitenfeld and P. H. Geubelle, Numerical anal- [S14] A. Molinari and H. Perfettini, Fundamental aspects of ysis of dynamic debonding under 2D in-plane and 3D a new micromechanical model of rate and state friction, loading, Int. J. Fract. 93, 13 (1998). J. Mech. Phys. Solids 124, 63 (2019). [S6] F. Barras, D. S. Kammer, P. H. Geubelle, and J.-F. [S15] M. Cocco and A. Bizzarri, On the slip-weakening be- Molinari, A study of frictional contact in dynamic frac- havior of rate- and state dependent constitutive laws, ture along bimaterial interfaces, Int. J. Fract. 189, 149 Geophys. Res. Lett. 29, 1516 (2002). (2014). [S16] A. Bizzarri and M. Cocco, Slip-weakening behavior dur- [S7] D. J. Andrews, Rupture propagation with nite stress ing the propagation of dynamic ruptures obeying rate- in antiplane strain, J. Geophys. Res. 81, 3575 (1976). and state-dependent friction laws, J. Geophys. Res. [S8] L. B. Freund, Dynamic Fracture Mechanics (Cambridge 108, 2373 (2003). university press, Cambridge, 1998). [S17] A. M. Rubin and J.-P. Ampuero, Earthquake nucleation [S9] E. Jones, T. Oliphant, and P. Peterson, SciPy: Open on (aging) rate and state faults, J. Geophys. Res. Solid source scienti c tools for Python (2001). Earth 110, B11312 (2005). http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Condensed Matter arXiv (Cornell University)

The emergence of crack-like behavior of frictional rupture: Edge singularity and energy balance

Loading next page...
 
/lp/arxiv-cornell-university/the-emergence-of-crack-like-behavior-of-frictional-rupture-edge-xYqXkqF4eb

References

References for this paper are not available at this time. We will be adding them shortly, thank you for your patience.

ISSN
0012-821X
eISSN
ARCH-3331
DOI
10.1016/j.epsl.2019.115978
Publisher site
See Article on Publisher Site

Abstract

The emergence of crack-like behavior of frictional rupture: Edge singularity and energy balance 1 2 1 Fabian Barras , Michael Aldam , Thibault Roch , E m 3;4 2  1y A. Brener , Eran Bouchbinder , and Jean-Fran cois Molinari Civil Engineering Institute, Materials Science and Engineering Institute, Ecole Polytechnique F ed erale de Lausanne, Station 18, CH-1015 Lausanne, Switzerland Chemical and Biological Physics Department, Weizmann Institute of Science, Rehovot 7610001, Israel Peter Grunb  erg Institut, Forschungszentrum Julich,  D-52425 Julich,  Germany Institute for Energy and Climate Research, Forschungszentrum Julich,  D-52425 Julich,  Germany The failure of frictional interfaces | the process of frictional rupture | is widely assumed to feature crack-like properties, with far-reaching implications for various disciplines, ranging from engineering tribology to earthquake physics. An important condition for the emergence of a crack- like behavior is the existence of stress drops in frictional rupture, whose basic physical origin has been recently elucidated. Here we show that for generic and realistic frictional constitutive relations, and once the necessary conditions for the emergence of an e ective crack-like behavior are met, frictional rupture dynamics are approximately described by a crack-like, fracture mechanics energy balance equation. This is achieved by independently calculating the intensity of the crack-like singularity along with its associated elastic energy ux into the rupture edge region, and the frictional dissipation in the edge region. We further show that while the fracture mechanics energy balance equation provides an approximate, yet quantitative, description of frictional rupture dynamics, interesting deviations from the ordinary crack-like framework | associated with non-edge-localized dissipation | exist. Together with the recent results about the emergence of stress drops in frictional rupture, this work o ers a comprehensive and basic understanding of why, how and to what extent frictional rupture might be viewed as an ordinary fracture process. Various implications are discussed. I. BACKGROUND AND MOTIVATION  =  , the di erence between the applied driving d res stress  and the residual stress  , in frictional rup- d res ture. In a very recent paper [1] we showed that, contrary Rapid slip along interfaces separating bodies in fric- to widely adopted assumptions, the residual stress res tional contact is mediated by the spatiotemporal dynam- is not a characteristic property of frictional interfaces. ics of frictional rupture [41, 43], which is a fundamental Rather, for rapid rupture  is shown to crucially de- res process of prime importance for a broad range of physi- pend on elastodynamic bulk e ects | in particular wave cal systems. For example, it is responsible for squealing radiation from the frictional interface to the bulks sur- in car brake pads [36], for bowing on a violin string [16], rounding it and long-range elastodynamic bulk interac- and for earthquakes along geological faults [8, 28, 35], tions | and that the existence of a nite stress drop  , to name just a few well-known examples. A very power- is a nite time e ect, limited by the wave travel time in ful conceptual and quantitative framework to understand nite systems. Speci cally, it has been shown that frictional dynamics in a wide variety of physical contexts is the analogy between frictional rupture and ordinary ( ) ' v ( ) ; (1) fracture/cracks. d d res 2c This framework is extensively used to interpret and quantify geophysical observations [2, 12], as well as a where  is the shear modulus of the bulks surrounding the broad spectrum of laboratory phenomena [7, 26, 27, 34, frictional interface, c is the corresponding shear wave- 40, 42, 44, 45]. For example, a recent series of careful lab- speed and v is the theoretically predicted residual slip res oratory experiments [7, 44, 45] demonstrated that when velocity behind the propagating rupture edge. v ( ) is res the analogy between frictional rupture and ordinary frac- determined through the approximate equation  (v ) + ss res ture holds, the dynamic propagation of laboratory earth- v ' , once long-range elastodynamic contributions res 2c quakes and their arrest can be quantitatively understood are omitted [1], where  (v) is the steady-state friction ss to an unprecedented degree [24]. Yet, the fundamental curve as a function of slip velocity v. physical origin and range of validity of the analogy be- The theoretical prediction in Eq. (1) has been sup- tween frictional rupture and ordinary fracture are not yet ported by existing experimental results for rapid fric- fully understood. tional rupture [1], for times shorter than the waves re- An important condition for the analogy to hold is ection time from outer boundaries, and by computer the emergence of a nite and well-de ned stress drop simulations in in nite systems. An example taken from one of these computer simulations is presented in Fig. 1a (cf. Fig. 3 in Barras et al. [1]), where two rapid rup- eran.bouchbinder@weizmann.ac.il ture fronts propagating in opposite directions are ob- jean-francois.molinari@ep .ch served, leaving behind them a well-de ned stress drop arXiv:1907.04376v2 [cond-mat.soft] 20 Nov 2019 2 that quantitatively agrees with the theoretical pre- rupture propagation is shown to be non-edge-localized, dictions (see Barras et al. [1] for details). The most out- i.e. to be spatially extended, and as such demonstrates standing theoretical question that remains open in the interesting deviations from the ordinary crack-like frame- context of the analogy between frictional rupture and or- work. Finally, the signi cance and implications of our dinary cracks, once the necessary conditions associated ndings for various phenomena are brie y discussed in with the emergence of a nite stress drop  are met, Sect. IV. Together with the recent results about the is to what extent the analogy actually holds, both in emergence of stress drops in frictional rupture [1], this qualitative and in quantitative terms. This question is work o ers a comprehensive and basic understanding of systematically addressed in this paper. why, how and to what extent frictional rupture might be viewed as an ordinary fracture process. The existence of a nite stress drop  does not im- mediately guarantee that the analogy between frictional rupture and ordinary fracture holds because proper scale separation should also be satis ed. That is, the resid- (a) ual stress  behind the propagating rupture should be res reached on a scale (typically termed the cohesive zone) 0.38 much smaller than the rupture size L (cf. Fig. 1a). If such scale separation is valid, we expect all crack-like properties to emerge in frictional rupture. In particu- 0.36 lar, we expect the frictional stress and slip velocity elds near the rupture edge to feature the famous square root singularity of conventional fracture mechanics [8]. More- 0.34 over, under these conditions, we expect the singularity- -4 -2 0 2 4 associated energy ux into the edge region to balance the edge-localized energy dissipation in excess of the power (b) 0.45 invested against the residual stress  . This energy bal- res ance relation amounts to an e ective equation of motion 0.4 for rupture propagation [8]. 0.35 In this paper we show that for generic and realistic frictional constitutive relations, and once the conditions 0.3 -7 -5 -3 -1 0 for the emergence of an e ective crack-like behavior are 10 10 10 10 10 met, frictional rupture dynamics are approximately | yet quantitatively { described by a crack-like, fracture mechanics energy balance equation [8]. This is achieved FIG. 1. (a) A snapshot of the frictional stress  (x) (normal- ized by the normal stress ) during rupture propagation that in a few steps. In Sect. II we show that if one assumes emerges in dynamic simulations with the steady-state friction the existence of the conventional square root singularity law shown in panel (b) and  = 0:375 (see text and Bar- of ordinary fracture mechanics and the associated near- ras et al. [1] for additional details). The snapshot reveals edge energy balance in frictional rupture, the latter fol- two rapid rupture fronts (the rupture length L is marked) lows a generic rupture length-velocity relation based on propagating at an instantaneous speed c ' 0:84c in oppo- r s the knowledge of the stress drop  alone. In Sect. III, site directions into regions characterized by the applied stress we quantitatively and systematically test these assump- and leaving behind them a well-de ned residual stress tions separately. We rst show that the conventional <  . Consequently, a well-de ned and nite stress drop res d square root singularity of standard fracture mechanics  emerges, as marked. Note that the y-axis is truncated provides a good quantitative description of the near rup- at = = 0:4 for visual clarity and that x is normalized by a generalized Grith-like length L , de ned in Eq. (6) (with ture edge stress and slip velocity elds simultaneously. a unity prefactor). (b) The steady-state friction stress  (v), ss We then propose a physically-motivated procedure to in- normalized by a constant normal stress , vs. the slip rate dependently extract an e ective fracture energy from the v (solid brown line). The curve has a generic N -shape [13], dissipative interfacial dynamics and show that it is bal- with a maximum at an extremely low v and a minimum at anced by the singularity-associated energy ux into the an intermediate v. The horizontal line represents the driving edge region to a good approximation. stress  , which intersects the N -shaped steady-state friction These results indicate that the scale separation men- curve at three points; the leftmost and rightmost ones are tioned above is approximately satis ed for frictional rup- stable xed points, while the intermediate one is an unstable one. The e ective steady-state friction curve (dash-dotted ture and that indeed the e ective fracture energy corre- orange line) is obtained by adding v (with  = 9GPa and sponds to edge-localized dissipation. However, the pro- 2c c = 2739m/s) to the solid brown line, see Barras et al. [1] for posed procedure to extract the relevant edge-localized s more details. The stress drop  of Eq. (1), which equals the dissipation allows us to show, also in Sect. III, that there one shown in panel (a), is marked by the black double-arrow. exists additional energy dissipation in excess of the power invested against the residual stress  . This contribu- res tion to the energy dissipation associated with frictional 3 II. CRACK-LIKE SCALING AND THE the energy release rate [23], even though it is not a rate), DEPENDENCE OF THE LENGTH-VELOCITY taking the form [23] RELATION ON THE STRESS DROP [K (L; c )] G(L; c )  A(c =c ) ; (4) r r s As explained above, and with the results of Barras et al. [1] in mind, we aim at carefully exploring the impli- where A(c =c ) is a known universal and dimensionless r s cations of stress drops | once they exist | for frictional function that depends on the fracture symmetry mode dynamics. The expected implications, to be detailed be- (here mode-II or mode-III). Finally, by invoking energy low, directly follow from the analogy to ordinary frac- balance in the edge region, standard fracture mechanics ture mechanics and consequently from its standard pre- predicts that [8] dictions [8, 43]. The challenge is to test whether these predictions are satis ed as emergent properties of the un- G(L; c ) = G (c ) ; (5) r c r derlying physics without assuming them a priori. Some of these predictions have been previously studied in the where G (c ) is the e ective fracture energy (of dimen- c r literature [10, 15{17, 19, 33, 46], but to the best of our sions of energy per unit area) associated with the transi- knowledge these studies have not yet led to a compre- tion from the v 0 state ahead of the edge to the v > 0 hensive picture of the analogy between frictional rupture state behind it, which possibly depends on the rupture and ordinary fracture. speed c . It is crucial to understand that unlike ordinary The existence of a stress drop behind the two edges tensile (mode-I symmetry) fracture, where G (c ) is the c r of propagating frictional rupture, cf. Fig. 1a, suggests only dissipation in the problem, in the friction problem that the load bearing capacity of the interface in this frictional dissipation exists everywhere along the sliding region is reduced,  < , and consequently that parts res d interface and not just in the transition region near the of the interface ahead of the edges should compensate for rupture edge. The way energy dissipation is partitioned this reduction, i.e. carry stress that is larger than  . In in the friction problem will be discussed below. the framework of the classical theory of fracture, the so- The above discussion raises several basic questions; called Linear Elastic Fracture Mechanics (LEFM), this most notably, does the square root singularity of Eqs. (2)- stress ampli cation ahead of the rupture edges follows a (3) generically exist in frictional rupture once  ex- universal singularity as the rupture edge is approached [8] ists? Can the e ective fracture energy G (c ) be mean- c r ingfully separated from the entire dissipation associated K (L; c ) (x)  p ; K (L; c )   LK(c =c ) ; (2) with frictional motion? And if so, can the energy bal- r r s jx x j ance of Eq. (5) be veri ed by independently calculating both G and G (the latter using Eq. (4))? While various where K quanti es the intensity of the singularity (hence aspects of these questions have certainly been addressed it is termed the stress intensity factor [23]), x is the lo- in the literature [10, 15{17, 19, 33, 46], we believe that cation of each of the rupture edges, L is the instanta- systematically addressing all of them in a single system is neous distance between the two edges (i.e. the rupture still missing. Before performing such a systematic anal- length/size, cf. Fig. 1a) and K(c =c ) is a dimension- r s ysis, we address rst a rather strong implication of the less function of the instantaneous propagation speed c of relations discussed above. each edge. We note that here and below numerical pre- Combining Eqs. (2)-(5), one obtains the following factors are omitted as we are interested in crack-like scal- stress drop dependent length-velocity relation ing relations in this section. In addition, the slip velocity is predicted to follow the very same singular behavior c =c = F [L=L ( )] with L ( )  ; (6) r s G G ( ) c K (L; c ) r r v(x)  p ; (3) jx x j which is valid under the assumption that G is indepen- dent of c . Here L ( ) is a generalized Grith-like r G just behind the edges (note the absolute value). As length [7, 8] and F () is a monotonically increasing func- expected, the intensity of the ampli cation/singularity tion that we do not specify. K (L; c ) in Eq. (2) increases with increasing  and the To test this prediction, we employed the generic rate- rupture length L (L is the size of the region in which and-state friction constitutive framework, presented in the interfacial load bearing capacity is reduced, hence a detail in Barras et al. [1]. Within this framework, the larger compensation/ampli cation exists). The relations interfacial constitutive law at any position x along the in Eqs. (2)-(3) are valid independently of the symme- interface and at any time t is described by the following try mode of rupture, and in particular in the context of local relation frictional rupture, they are valid for both in-plane shear (mode-II) and anti-plane shear (mode-III) symmetries. =  sgn(v) f (jvj; ) ; (7) Standard fracture mechanics predicts that the square root singularity in Eqs. (2)-(3) is accompanied by a nite which must be supplemented with a dynamical equation ux of energy G into the rupture edge region (known as for the evolution of . Extensive evidence indicates that 4 1 f (jvj;  = D=v) =  (v)=, under steady-state sliding ss ss conditions and a controlled normal stress , has been measured over a broad range of slip rates v for many materials [11]. 0.9 Together with general theoretical considerations [13], it is now established that the steady-state frictional stress (v) is generically N -shaped, as shown in Fig. 1b (solid ss brown line). Finally, the e ective friction curve obtained by adding the radiation damping term v, which has 0.8 2c been shown to play an important role in the emergence of stress drops in frictional rupture [1], is also presented in Fig. 1b (dash-dotted orange line). We would like to 0.5 0.6 0.7 0.8 stress that, as shown in Barras et al. [1], pure velocity- weakening friction laws also e ectively feature N -shaped behavior due to the radiation damping term (and hence also feature a nite stress drop). Consequently, the re- sults to be presented below equally apply to velocity- weakening friction laws. 0.9 Coupling this constitutive framework to spectral boundary integral method [5, 22, 30] calculations in in- nite systems under mode-III deformation conditions, gave rise to frictional rupture such as the one shown 0.8 in Fig. 1a. In this approach, the displacement eld ^ ^ ^ ^ u(x; y; t) =u (x; y; t)z (the unit vectors satisfy z? x; y) is computed at the interface y! 0 self-consistently with 5 10 15 20 the far- eld stress  and the friction law of Eq. (7), see Barras et al. [1] for additional details. Based on such numerical computations, we plot in Fig. 2a the nor- FIG. 2. (a) The frictional rupture velocity c , normalized by malized frictional rupture velocity c =c vs. the frictional r r s c , as a function of the frictional rupture length L (normal- s rupture length L for various driving stress levels  (de- ized by the system size W = 320m used in these calculations) tailed in the legend of Fig. 2b). The di erent c (L) curves for di erent driving stress levels  = (as detailed in the leg- span a rather broad range. Equation (6) predicts that end of panel (b)), using the N -shaped friction law of Fig. 1b. these curves can be collapsed onto a master curve if L is Frictional rupture is nucleated as described in Barras et al. rescaled by L ( ), where  ( ) is given in Eq. (1) (see G d [1]. (b) The prediction of Eq. (6) is tested by plotting c =c r s also Fig. 3c in Barras et al. [1]) and the e ective fracture vs. L=L ( ), where  varies with  according to Eq. (1) G d energy G is assumed to be independent of c . To follow c r (see also Fig. 3c in Barras et al. [1]). L ( ), as de ned in this rescaling procedure, L ( ) of Eq. (6) is evaluated Eq. (6), is evaluated with  = 9GPa, G = 0:65J/m and a with a unity prefactor,  = 9GPa and G = 0:65J/m . unity prefactor. The length-velocity curves of panel (a) all c The way to extract the value of the e ective fracture en- collapse on a master envelope curve as predicted by Eq. (6), see additional discussion in the text. ergy G is discussed in Sect. III below. The outcome of the rescaling procedure is presented in Fig. 2b. physically represents the age/maturity of the contact (hence it is related to the real contact area) [9, 11, 20, It is observed that the di erent c (L) curves, which ex- 29, 31, 32, 38], and that its evolution takes the form hibited a rather large spread in Fig. 2a, collapse on the envelope of a single master curve upon rescaling L by jvj L ( ). Note that deviations from the master curve are = g ; (8) observed at early times (small L values in each curve); this is expected as the crack-like behavior cannot be valid with g(1) = 0 and where  is of time dimension. The in the nucleation stage, but rather only when L is suf- characteristic slip displacement D controls the transition ciently large and frictional rupture is suciently well- from a stick state v 0, with a characteristic structural developed. The collapse in Fig. 2b provides indirect, yet state  =  , to a steadily slipping/sliding state v > 0, strong, support to the applicability of the crack-like rela- with  =D=v. The precise functional form of g() (with tions in Eqs. (2)-(5) to frictional rupture. These relations ss g(1) = 0) plays no role in what follows. The function will be directly tested next. 5 III. THE EMERGENCE OF STRESS for the di erent locations x's at small , but exhibits lo- SINGULARITY AND LOCAL ENERGY cation dependence at signi cantly larger , where it levels BALANCE o to di erent limiting values that become closer to one another as x increases. These observations can be under- stood as follows; the frictional stress  (x; t) presented in One of the major implications of the existence of a - Fig. 3a exhibits two distinct behaviors behind the prop- nite stress drop  is the emergence of stress singularity agating rupture edge (here the propagation is from right near the frictional rupture edge, as explained above and to left). First, it features a strong decay well within the as formulated in Eqs. (2)-(3). In order to directly test this edge region. Second, as denoted by the arrow, there ex- prediction, we present in Fig. 3a the (properly normal- ists a transition to a slow decay towards  on a signi - ized) spatial pro les of  (x; t) and v(x; t) near a rupture res cantly larger lengthscale, extending far beyond the edge edge at time t. We then t the two elds together to region (the full spatial extent of this decay is not shown). Eqs. (2)-(3), demanding the same stress intensity factor This slow spatial decay stems from the rate and state de- K and the same e ective tip location x (the details of pendence of the friction law, which implies that all of the tting procedure are extensively discussed in the SM the interfacial elds in the problem  (x; t); v(x; t); (x; t) [1]). slowly approach their respective asymptotic steady-state The resulting ts are superimposed on the elds  (x; t) values  ; v ; D=v . Finally, as rupture propagation res res res and v(x; t) in Fig. 3a. The square root singular behav- in the presence of a nite stress drop is intrinsically out ior faithfully describes the two elds near the front edge, of steady state, i.e. rupture accelerates towards c as supporting the prediction that such a singular behavior shown in Fig. 2, we expect some position dependence emerges in the presence of a nite stress drop  . Note of E (; x). This dependence should become weaker as BD that the spatial range in which the elds are described by the limiting velocity c ! c is approached, as is indeed the square root singular behavior is larger for the slip ve- r s observed in Fig. 3b. locity v(x; t) than for the frictional stress  (x; t). The rea- The physical picture emerging from the above discus- son is that  (x; t) features a signi cantly narrower range sion suggests that the location independent part of the of values between its peak value and the applied stress breakdown energy E (; x), which is associated with BD (in the large jxj limit) compared to the corresponding excess dissipation near the rupture edge, should be iden- range for v(x; t), and thus the latter can accommodate a ti ed as the e ective fracture energy G appearing in singular behavior, which is by construction an interme- Eq. (5). This idea is pictorially demonstrated by the diate asymptotic behavior, over a larger spatial range. horizontal black line in Fig. 3b, which identi es G with The results of Fig. 3a demonstrate that a rather well- the point in which the various E (; x) curves start to BD de ned stress intensity factor K (L; c ) is associated with split/deviate one from another (from which a value of frictional rupture in the presence of a nite stress drop G  0:65J/m can be inferred). To make the identi ca- , from which the energy release rate G(L; c ) can be tion of G more quantitative and to allow a direct test readily extracted using Eq. (4) [1]. Next, in order to test of Eq. (5), we invoke the observation that the combina- the validity of Eq. (5), we need to independently calculate tion v=D strongly overshoots unity in the edge region the e ective fracture energy G associated with frictional (v=D> 1 implies < 0, which is associated with contact rupture propagation. To this aim, we de ne the energy area reduction), then slightly undershoots it and nally per unit area that is dissipated at a given interfacial loca- approaches unity from below far from the edge [1]. We tion x during the transition from a non-slipping/sticking note that the position of the rst crossing v=D = 1 ap- state to a steadily sliding state characterized by the resid- proximately corresponds to the position marked by small ual stress  [10] res arrow in Fig. 3a. Consequently, the edge-localized dis- sipation G can be estimated as the excess dissipation 0 0 E (; x) =  ( )  d : (9) BD res associated with the spatial region for which v=D > 1, quanti ed by the following spatial integral Here the slip history at a location x is given by the slip G (c )   (x; t)  v(x; t) dx : (10) displacement (x; t)  u (x; y = 0 ; t)u (x; y = 0 ; t), c r res z z c (t) r v=D>1 where (x; t) =v(x; t), and the subscript 'BD' stands for 'breakdown'. The breakdown energy quanti es the excess We note that this estimate of G appears to be consis- dissipation on top of the frictional dissipation associated tent with an analytic approximation available in the lit- with sliding against the residual stress  . Note that we erature [15{17], which may shed light on the dependence res cannot a priori identify the breakdown energy de ned in of G on interfacial parameters (see SM [1] for details). Eq. (9) with the e ective fracture energy G , as will be We are now in a position to directly test Eq. (5), where discussed next. the energy release rate G is calculated using the stress In Fig. 3b we plot the breakdown energy E (; x) at intensity factor extracted as shown in Fig. 3a and G BD c 4 di erent interfacial locations x =` , i = 14, ordered by through Eq. (10). In the inset of Fig. 3b, we plot the their proximity to the nucleation site (the center of the ratio G=G as a function of the rupture length L. It is domain). It is observed that E (; x) perfectly overlaps observed that G=G is close to unity throughout the rup- BD c 6 0.25 1.2 -1 0.2 0.8 -2 0.15 -4 -3 -2 10 10 10 1.25 0.6 0.1 0.4 0.75 0.05 0.2 0.5 0.2 0.4 0.6 0.8 0 0 -0.004 -0.002 0 0.002 0.004 0 0.1 0.2 0.3 0.4 0.5 FIG. 3. (a) The normalized spatial pro les of  (x; t) and v(x; t) near a rupture edge propagating from right to left with a velocity c ' 0:94c at time t. x is shifted by x , which corresponds to the location of e ective rupture edge (cf. Eqs. (2)-(3)). r s r 2 2 Both elds are normalized/shifted by quantities de ned in the text, except for  1 c =c . The dashed lines are the results r s 1=2 of tting the solid lines to Eqs. (2)-(3), with K = 64kPam , see SM [1] for additional details. The tilted arrow is discussed in the text. (inset) The same as the main panel, but on a double logarithmic scale with the x-axis being jx x j. Note that since the dashed lines in the main panel are symmetric with respect to x , using jxx j implies the existence of a single dashed r r line in the inset. The inset highlights both the quality of the t and the di erent spatial ranges used for each eld, see SM [1] for additional details. (b) The breakdown energy E (; x), de ned in Eq. (9), vs. slip  for 4 interfacial locations x =` , with BD i ` =W = 0:15, ` =W = 0:20, ` =W = 0:25 and ` =W = 0:30. ` are measured from the nucleation site (the center of the system) 1 2 3 4 i and the system size is W = 80m. The horizontal black line marks the splitting of the di erent curves, which is identi ed with G  0:65J/m . (inset) G=G vs. L=W , where L is the rupture length. G is calculated using K (L), cf. panel (a) and the SM c c [1], through Eq. (4) and G is calculated through Eq. (10). The generic properties of the results presented in this gure are independent of the details of the friction law (not shown). ture propagation history, lending strong support to the IV. SUMMARY AND CONCLUDING REMARKS ideas developed above. In particular, it shows that the rupture edge energy balance in Eq. (5) provides quanti- tative approximations for frictional rupture dynamics. In this paper we set out to further explore the analogy between frictional rupture and ordinary fracture. The At the same time, our results also clearly demon- starting point for this investigation is our own very recent strate that E (; x) can be quite signi cantly larger BD work that elucidated the physical origin of stress drops than G and position dependent, implying that non- in frictional rupture [1], which constitute a necessary edge-localized dissipation in excess of the power invested condition for the analogy. Our major goal was to under- against the residual stress  is a generic property of res stand to what extent the analogy holds, both in quali- frictional interfaces featuring rate and state dependent tative and in quantitative terms, for interfaces described friction. A similar physical situation has been discussed by generic and realistic frictional constitutive relations, in Brener and Marchenko [15]. That is, while a physi- once stress drops do exist. cally sensible extraction of the edge-localized excess dis- sipation G allows to obtain reasonably well quantitative We showed that for rate and state constitutive rela- approximations for frictional rupture dynamics based on tions, frictional rupture dynamics are approximately | the analogy to ordinary fracture, our results clearly in- yet quantitatively | described by an ordinary fracture dicate that this analogy is incomplete and that interest- energy balance equation, when the conditions for the ing deviations exist. These deviations are intimately re- emergence of a nite stress drop  are satis ed. To es- lated to the spatially extended (non-edge-localized) rate tablish the quantitative status of this fracture mechanics and state dependence of frictional interfaces, an intrin- energy balance equation, we proposed a physical criterion sic frictional property that is entirely absent in ordinary for extracting the rupture edge-localized dissipation di- fracture, and are manifested in non-edge-localized excess rectly from the frictional dynamics, allowing to de ne an dissipation. The latter may have important implications e ective fracture energy G for frictional problems. Sur- for the energy budget associated with frictional dynam- prisingly, we discovered that G does not account for all ics, and might be relevant to geophysical observations of the energy dissipation E in excess of the energy dis- BD and their interpretations [17, 19, 33, 46]. sipated against the residual stress  (cf. Eq. (9)). These res 7 ndings imply that the analogy between frictional rup- et al. [17] and Tinti et al. [46], which concluded based on ture and ordinary fracture is not complete, as manifested seismic data that the breakdown energy can be larger by the existence of a non-edge-localized contribution to than the fracture energy for large earthquake ruptures. E . These results o er insight into open questions concerning BD earthquake energy budget [2, 17, 19, 33, 46] and deserve The di erence between E and G is intimately re- additional investigation. BD c lated to the generic rate and state dependence of friction, More generally, we expect our results to provide which is responsible for the two-step nature of the stress a conceptual and quantitative framework to address relaxation/weakening process associated with frictional various fundamental and applied problems in relation rupture propagation; rst, there exists a rather sharp to the rupture dynamics of frictional interfaces, with stress drop that takes place over a relatively small slip, implications for both laboratory and geophysical-scale bringing the stress close to, but not identically to, the phenomena. For example, our results and theoretical residual stress  . Second, there exists a slower, longer- framework are expected to apply also to slip pulses. res term process that brings the stress to the residual stress Indeed, recent preliminary results, see Fig. S6 in Brener over signi cantly larger slip. The latter stress relax- et al. [2], support this expectation. res ation/weakening process, which some authors attribute to melting or thermal pressurization [37, 47] not taken Acknowledgements E. B. and J.-F.M. acknowledge sup- into account in the present work, is responsible for the port from the Rothschild Caesarea Foundation. E. B. ac- di erence between E and G . This physical picture knowledges support from the Israel Science Foundation BD c is reminiscent of the model proposed in Kanamori and (Grant No. 295/16). J.-F.M., F. B. and T. R. acknowl- Heaton [25], and further discussed in Abercrombie and edge support from the Swiss National Science Founda- Rice [2], in trying to resolve some puzzling observations tion (Grant No. 162569). This research is made possible in relation to the energy budget of earthquake rupture. in part by the historic generosity of the Harold Perlman Moreover, this physical picture is consistent with Chester Family. [1] See Supplemental Material for additional information, . progressive evolutionary changes, and di erent dy- [2] Abercrombie, R.E., Rice, J.R.. Can observa- namic regimes. Rev Geophys 2008;46(4):RG4006. tions of earthquake scaling constrain slip weaken- URL: http://doi.wiley.com/10.1029/2008RG000260. ing? Geophys J Int 2005;162(2):406{424. URL: doi:doi:10.1029/2008RG000260. https://academic.oup.com/gji/article-lookup/ [9] Bhattacharya, P., Rubin, A.M.. Frictional re- doi/10.1111/j.1365-246X.2005.02579.x. doi:doi: sponse to velocity steps and 1-D fault nucleation un- 10.1111/j.1365-246X.2005.02579.x. der a state evolution law with stressing-rate dependence. [7] Andrews, D.J.. Rupture propagation with nite stress J Geophys Res Solid Earth 2014;119(3):2272{2304. in antiplane strain. J Geophys Res 1976;81(20):3575{ URL: http://doi.wiley.com/10.1002/2013JB010671. 3582. URL: http://doi.wiley.com/10.1029/ doi:doi:10.1002/2013JB010671. JB081i020p03575. doi:doi:10.1029/JB081i020p03575. [10] Bizzarri, A.. On the relations between fracture [13] Bar-Sinai, Y., Spatschek, R., Brener, E.A., Bouch- energy and physical observables in dynamic earth- binder, E.. On the velocity-strengthening behav- quake models. J Geophys Res 2010;115(B10):B10307. ior of dry friction. J Geophys Res Solid Earth URL: http://doi.wiley.com/10.1029/2009JB007027. 2014;119(3):1738{1748. URL: http://doi.wiley.com/ doi:doi:10.1029/2009JB007027. 10.1002/2013JB010586. doi:doi:10.1002/2013JB010586. [16] Bizzarri, A., Cocco, M.. Slip-weakening behavior [1] Barras, F., Aldam, M., Roch, T., Brener, E.A., Bouch- during the propagation of dynamic ruptures obeying binder, E., Molinari, J.F.. The emergence of crack- rate- and state-dependent friction laws. J Geophys Res like behavior of frictional rupture: The origin of stress 2003;108(B8):2373. URL: http://doi.wiley.com/10. drops. To appear in Physical Review X 2019. URL: http: 1029/2002JB002198. doi:doi:10.1029/2002JB002198. //arxiv.org/abs/1906.11533. arXiv:1906.11533. [12] Bizzarri, A., Liu, C.. Near- eld radiated wave eld [11] Baumberger, T., Caroli, C.. Solid friction from may help to understand the style of the supershear stickslip down to pinning and aging. Adv Phys transition of dynamic ruptures. Phys Earth Planet 2006;55(3-4):279{348. URL: http://www.tandfonline. Inter 2016;261:133{140. URL: https://linkinghub. com/doi/abs/10.1080/00018730600732186. doi:doi: elsevier.com/retrieve/pii/S0031920116300838. 10.1080/00018730600732186. doi:doi:10.1016/j.pepi.2016.05.013. [7] Bayart, E., Svetlizky, I., Fineberg, J.. Frac- [5] Breitenfeld, M.S., Geubelle, P.H.. Numeri- ture mechanics determine the lengths of interface rup- cal analysis of dynamic debonding under 2D in- tures that mediate frictional motion. Nat Phys plane and 3D loading. Int J Fract 1998;93(1/4):13{ 2015;12(2):166{170. URL: https://www.nature.com/ 38. URL: http://link.springer.com/10.1023/A: articles/nphys3539. doi:doi:10.1038/nphys3539. 1007535703095. doi:doi:10.1023/A:1007535703095. [8] Ben-Zion, Y.. Collective behavior of earth- [2] Brener, E.A., Aldam, M., Barras, F., Molinari, J.F., quakes and faults: Continuum-discrete transitions, Bouchbinder, E.. Unstable Slip Pulses and Earthquake 8 Nucleation as a Nonequilibrium First-Order Phase 2010;115(12):1{25. doi:doi:10.1029/2009JB006833. Transition. Phys Rev Lett 2018;121(23):234302. URL: [28] Marone, C.. Laboratoty-derived friction https://link.aps.org/doi/10.1103/PhysRevLett. laws and their application to seismic faulting. 121.234302. doi:doi:10.1103/PhysRevLett.121.234302. Annu Rev Earth Planet Sci 1998;26(1):643{ [15] Brener, E.A., Marchenko, V.I.. Frictional shear 696. URL: http://www.annualreviews.org/doi/ cracks. J Exp Theor Phys Lett 2002;76(4):211{214. abs/10.1146/annurev.earth.26.1.643. doi:doi: URL: http://link.springer.com/10.1134/1.1517386. 10.1146/annurev.earth.26.1.643. doi:doi:10.1134/1.1517386. [29] Marone, C.. The e ect of loading rate on [16] Casado, S.. Studying friction while playing the vi- static friction and the rate of fault healing during olin: exploring the stickslip phenomenon. Beilstein the earthquake cycle. Nature 1998;391(6662):69{72. J Nanotechnol 2017;8:159{166. URL: http://www. URL: http://www.nature.com/nature/journal/v391/ beilstein-journals.org/bjnano/content/8/1/16. n6662/abs/391069a0.html. doi:doi:10.1038/34157. doi:doi:10.3762/bjnano.8.16. [30] Morrissey, J.W., Geubelle, P.H.. A numerical scheme [17] Chester, J.S., Chester, F.M., Kronenberg, A.K.. for mode III dynamic fracture problems. Int J Numer Fracture surface energy of the Punchbowl fault, San Methods Eng 1997;40(7):1181{1196. URL: https: Andreas system. Nature 2005;437(7055):133{136. //doi.org/10.1002/(SICI)1097-0207(19970415)40: URL: http://www.nature.com/articles/nature03942. 7{%}3C1181::AID-NME108{%}3E3.0.CO;2-X. doi:doi: doi:doi:10.1038/nature03942. 10.1002/(SICI)1097-0207(19970415)40:7<1181::AID- [15] Cocco, M., Bizzarri, A.. On the slip- NME108>3.0.CO;2-X. weakening behavior of rate- and state dependent con- [31] Nagata, K., Nakatani, M., Yoshida, S.. A re- stitutive laws. Geophys Res Lett 2002;29(11):1516. vised rate- and state-dependent friction law obtained URL: http://doi.wiley.com/10.1029/2001GL013999. by constraining constitutive and evolution laws sepa- doi:doi:10.1029/2001GL013999. rately with laboratory data. J Geophys Res Solid Earth [19] Das, S.. Dynamic fracture mechanics in the study of the 2012;117(B2):B02314. URL: http://doi.wiley.com/ earthquake rupturing process: theory and observation. 10.1029/2011JB008818. doi:doi:10.1029/2011JB008818. J Mech Phys Solids 2003;51(11-12):1939{1955. URL: [32] Nakatani, M.. Conceptual and physical clari - https://linkinghub.elsevier.com/retrieve/pii/ cation of rate and state friction: Frictional slid- S0022509603001443. doi:doi:10.1016/j.jmps.2003.09.025. ing as a thermally activated rheology. J Geo- [20] Dieterich, J.H.. Applications of rate-and state- phys Res Solid Earth 2001;106(B7):13347{13380. dependent friction to models of fault slip and earthquake URL: http://doi.wiley.com/10.1029/2000JB900453. occurrence. Treatise Geophys 2007;4:107{129. doi:doi: doi:doi:10.1029/2000JB900453. 10.1073/pnas.93.9.3787. [33] Nielsen, S.B., Spagnuolo, E., Smith, S.A.F., Vio- [8] Freund, L.B.. Dynamic Fracture Mechanics. Cambridge: lay, M., Di Toro, G., Bistacchi, A.. Scaling in nat- Cambridge university press, 1998. ural and laboratory earthquakes. Geophys Res Lett [22] Geubelle, P., Rice, J.R.. A spectral method 2016;43(4):1504{1510. URL: http://doi.wiley.com/ for three-dimensional elastodynamic fracture prob- 10.1002/2015GL067490. doi:doi:10.1002/2015GL067490. lems. J Mech Phys Solids 1995;43(11):1791{1824. [34] Noda, H., Lapusta, N., Kanamori, H.. Com- URL: http://linkinghub.elsevier.com/retrieve/ parison of average stress drop measures for ruptures pii/002250969500043I. doi:doi:10.1016/0022- with heterogeneous stress change and implications for 5096(95)00043-I. earthquake physics. Geophys J Int 2013;193(3):1691{ [23] Irwin, G.R.. Analysis of stresses and strains near the end 1712. URL: http://gji.oxfordjournals.org/cgi/doi/ of a crack traversing a plate. J Appl Mech 1957;24:361{ 10.1093/gji/ggt074. doi:doi:10.1093/gji/ggt074. 364. [35] Ohnaka, M.. The physics of rock failure and earthquakes. [24] Kammer, D.S., Radiguet, M., Ampuero, J.P., Moli- Cambridge University Press, 2013. nari, J.F.. Linear Elastic Fracture Mechanics Predicts [36] Rhee, S., Jacko, M., Tsang, P.. The role of friction the Propagation Distance of Frictional Slip. Tribol Lett lm in friction, wear and noise of automotive brakes. 2015;57(3):23. URL: http://link.springer.com/10. Wear 1991;146(1):89{97. URL: http://linkinghub. 1007/s11249-014-0451-8. doi:doi:10.1007/s11249-014- elsevier.com/retrieve/pii/004316489190226K. 0451-8. arXiv:1408.4413. doi:doi:10.1016/0043-1648(91)90226-K. [25] Kanamori, H., Heaton, T.H.. Microscopic and [37] Rice, J.R.. Heating and weakening of faults dur- macroscopic physics of earthquakes. In: Geocomplex- ing earthquake slip. J Geophys Res Solid Earth ity Phys. Earthquakes. American Geophysical Union 2006;111(B5):B05311. URL: http://doi.wiley.com/ (AGU); 2000. p. 147{163. URL: http://www.agu. 10.1029/2005JB004006. doi:doi:10.1029/2005JB004006. org/books/gm/v120/GM120p0147/GM120p0147.shtml. [38] Rice, J.R., Ruina, A.L.. Stability of Steady Frictional doi:doi:10.1029/GM120p0147. Slipping. J Appl Mech 1983;50(2):343{349. URL: [26] Lu, X., Lapusta, N., Rosakis, A.J.. Pulse- http://appliedmechanics.asmedigitalcollection. like and crack-like dynamic shear ruptures on fric- asme.org/article.aspx?articleid=1406945. doi:doi: tional interfaces: experimental evidence, numerical 10.1115/1.3167042. modeling, and implications. Int J Fract 2010;163(1- [17] Rubin, A.M., Ampuero, J.P.. Earthquake nucleation on 2):27{39. URL: http://link.springer.com/10.1007/ (aging) rate and state faults. J Geophys Res Solid Earth s10704-010-9479-4. doi:doi:10.1007/s10704-010-9479-4. 2005;110(B11):B11312. URL: http://doi.wiley.com/ [27] Lu, X., Rosakis, A.J., Lapusta, N.. Rupture 10.1029/2005JB003686. doi:doi:10.1029/2005JB003686. modes in laboratory earthquakes: E ect of fault prestress [40] Rubino, V., Rosakis, A.J., Lapusta, N.. Understanding and nucleation conditions. J Geophys Res Solid Earth dynamic friction through spontaneously evolving labo- 9 ratory earthquakes. Nat Commun 2017;8(7260):15991. URL: http://www.nature.com/doifinder/10.1038/ ncomms15991. doi:doi:10.1038/ncomms15991. [41] Scholz, C.H.. The mechanics of earthquakes and faulting. Cambridge university press, 2002. [42] Svetlizky, I., Bayart, E., Cohen, G., Fineberg, J.. Frictional Resistance within the Wake of Frictional Rup- ture Fronts. Phys Rev Lett 2017;118(23):234301. URL: http://link.aps.org/doi/10.1103/PhysRevLett.118. 234301. doi:doi:10.1103/PhysRevLett.118.234301. [43] Svetlizky, I., Bayart, E., Fineberg, J.. Brittle Frac- ture Theory Describes the Onset of Frictional Motion. Annu Rev Condens Matter Phys 2019;10(1):031218{ 013327. URL: https://www.annualreviews.org/doi/ 10.1146/annurev-conmatphys-031218-013327. doi:doi: 10.1146/annurev-conmatphys-031218-013327. [44] Svetlizky, I., Fineberg, J.. Classical shear cracks drive the onset of dry frictional mo- tion. Nature 2014;509(7499):205{208. URL: https://www.nature.com/articles/nature13202. doi:doi:10.1038/nature13202. [45] Svetlizky, I., Pino Mun ~oz, D., Radiguet, M., Kam- mer, D.S., Molinari, J.F., Fineberg, J.. Prop- erties of the shear stress peak radiated ahead of rapidly accelerating rupture fronts that mediate fric- tional slip. Proc Natl Acad Sci 2016;113(3):542{ 547. URL: http://www.pnas.org/lookup/doi/10.1073/ pnas.1517545113. doi:doi:10.1073/pnas.1517545113. [46] Tinti, E., Spudich, P., Cocco, M.. Earth- quake fracture energy inferred from kinematic rup- ture models on extended faults. J Geophys Res 2005;110(B12):B12303. URL: http://doi.wiley.com/ 10.1029/2005JB003644. doi:doi:10.1029/2005JB003644. [47] Viesca, R.C., Garagash, D.I.. Ubiquitous weaken- ing of faults due to thermal pressurization. Nat Geosci 2015;8(11):875{879. URL: https://www.nature.com/ articles/ngeo2554. doi:doi:10.1038/ngeo2554. S1 Supplemental Material for: \The emergence of crack-like behavior of frictional rupture: Edge singularity and energy balance" The goal of this document is to provide additional tech- nical details regarding the extraction of the near-edge singular elds (Fig. 3a in the manuscript) and the ef- 0.5 fective fracture energy G from the interfacial dynamics (Fig. 3b in the manuscript), both discussed in Sect. III of the manuscript. This is achieved in two steps; rst, in 2 4 6 8 10 Sect. S-1, some relevant concepts and methodology are being discussed and tested using a conventional cohesive zone model of ordinary fracture. Then, in Sect. S-2, these concepts and tools are generalized for frictional rupture along interfaces described by generic friction constitutive relations, and additional details about their application -4 -3 -2 -1 0 1 2 3 4 in Sect. III of the manuscript are brie y provided. The numerical tools and the generic interfacial constitutive re- lation (including the material parameters) are presented FIG. S1. Space-time diagram of the dynamic mode-III rup- in [S1, S2]. ture event described in the text. The yellow region corre- sponds to the broken interface left behind the propagating rupture edges, the narrow red region corresponds to the co- S-1. EDGE SINGULARITY AND ENERGY hesive zone and the black region corresponds to the intact interface. The blue line marks the instant at which the snap- BALANCE IN A CONVENTIONAL COHESIVE ZONE MODEL OF ORDINARY FRACTURE shots of the stress and slip velocity elds in Fig. S2a are taken. (inset) The time evolution of the rupture speed c as function of its size L. Our goal here is to rst develop the procedure for extracting the near-edge singular elds in a simpler case, where there is no residual stress (i.e. ordinary the center of an interface at rest under a uniform shear fracture), where the Linear Elastic Fracture Mechanics stress  , where 0 <  <  , by progressively increasing d d c (LEFM) singularity is regularized on a small lengthscale an originally in nitesimal seed crack toward a critical (i.e. proper scale separation is realized) and the frac- size L = L . The latter, known as the Grith criti- ture energy G is prescribed. This is achieved by the cal length [S7, S8], is given by (see also Eq. (6) in the well-known framework of cohesive zone crack models, at- manuscript) tributed to Dugdale [S3] and Barenblatt [S4], which be- came very popular in the numerical modeling of dynamic 4 G L = ; (S3) fracture (see, for example, [S5, S6]). Within this frame- G work, we employ a linear slip-weakening cohesive law in str which the strength of the interface  linearly reduces for mode-III cracks. In Fig. S1, we present the resulting to zero over a characteristic slip displacement dynamics that feature a crack that progressively accel- erates toward c , the maximal admissible rupture speed str s (x; t) =  f1 (x; t)= g ; (S1) c c for mode-III symmetry. The instantaneous rate of dissipated energy associated where  is the failure strength (determining the rupture with the propagation of one rupture edge (recall that peak stress), (x; t) is the slip displacement, and fg = there are two of these) can be obtained as [S6] if  > 0 and 0 otherwise ( is a dummy variable used to de ne the function fg in Eq. (S1)). The linear slip- Z 1 weakening law of Eq. (S1) corresponds to a prescribed E (t) =  (x; t) v(x; t) dx ; (S4) diss value of the fracture energy 1 where W is the system size. The integral attains a nite G = d =   : (S2) c c c contribution only inside the well-de ned cohesive zone near the propagating rupture edge, where both  (x; t) The spectral boundary integral method under mode- and v(x; t) are non-zero. The cohesive zone (also termed III symmetry (where the basic object is the out-of-plane fracture process zone in ordinary fracture), which cor- displacement eld at the interface, u (x; y = 0; t), see responds to the region where the stress  (x; t) drops manuscript and references therein for details) can be cou- from the peak stress (failure strength)  to 0, is marked pled to Eq. (S1) (i.e. the latter replaces the friction law by the red-shaded region in Fig. S2a. A snapshot of used in the manuscript) to generate propagating rup- the stress  (x; t) and slip velocity v(x; t) distributions ture fronts. In this context, rupture is nucleated at near the propagating rupture edge are also presented in S2 Fig. S2a (and see also Fig. S1). The fracture energy, de- in Fig. S2a. That is, we t the normalized and shifted ned in Eq. (S2), is the energy dissipated per unit crack near-edge stress and slip velocity elds to the singular extension dL form in Eqs. (S6)-(S7), with x and K as the two free r III parameters. To make the procedure well de ned, we also d dE E (t) dL diss diss need to specify the spatial range over which the ts are G (t) = E (t) = = ; (S5) c diss dL dt dt c (t) r performed. In determining the spatial range of the t of the two elds, several physical considerations are invoked; which is constant for the slip-weakening model used here rst, it is clear that the ts cannot include the regions (see Fig. S2b). where the elds (cf. the examples in Fig. S2a) attain their Standard fracture theory predicts that close to the peak values as these are associated with the regulariza- propagating rupture edges, we have the famous square tion of the singular behavior (the cohesive zone). Second, root singular elds [S8] the tting ranges cannot extend too far away from the edge region as the elds there include also non-singular III (r =x x;  = 0; c )  ' p (S6) r r res contributions. Finally, as the overall variability of the 2(x x) stress eld is smaller compared to that of the slip veloc- ity eld, we expect the singular region to be narrower for and the former. We employ a nonlinear least-squares regres- (c ) K s r III sion tting procedure [S9] to determine the best estimates v(r =xx ;  =; c ) ' p ; (S7) r r 2c 2(x x ) for x and K , and selected the tting ranges to be as r r III large as possible within the constraints imposed by the where (r; ) is a polar coordinate system moving with physical considerations just stated. 2 2 the rupture edge, (c ) = 1 c =c , x is the e ective s r r r s edge location and K is the mode-III stress intensity III factor. We subtracted the residual stress  from the res The resulting ts, i.e. the right-hand-sides of Eqs. (S6)- frictional stress eld such that the shifted stress eld (S7), are superimposed on the normalized slip velocity v vanishes behind the rupture edge and normalized the and frictional stress  elds in Fig. S2a (dashed lines). slip velocity eld such that the left-hand-sides of both To highlight the spatial tting ranges used, we replot the Eqs. (S6)-(S7) attain comparable values; note that for results in Fig. S2a on a double logarithmic scale against the slip-weakening model used here we have  = 0, res jx x j=W in the inset (note that due to the symmetry and it makes no di erence, but in general one may have r of the singular form on the right-hand-sides of Eqs. (S6)- > 0 (also in the framework of slip-weakening models), res (S7), we have now a single t that describes the two elds see Sect. S-2. In addition, we used v = 2u _ since v is the over di erent spatial ranges). The inset shows that the slip velocity, not the particle (mass) velocity u _ . Finally, spatial tting ranges for the two elds are di erent, that as is evident from the right-hand-sides of both Eqs. (S6)- the range for the slip velocity eld is wider than the one (S7), the normalized slip velocity v and frictional stress for the frictional stress eld and that the peak regions are elds are symmetric functions relative to x (i.e. it is the properly excluded. Finally, we veri ed that the values very same function of jxx j), though the spatial ranges of x and K are robust against changes in the spatial r III in which the singular form is valid di er for the two elds. tting ranges within the stated constraints. This issue will be discussed below, where we explain how the two free parameters in Eqs. (S6)-(S7) | x and K r III | are determined. We stress that the proper normaliza- tion and shift used in Eqs. (S6)-(S7) allow us to consider The extracted value of K has been used to calculate III the stress and slip velocity elds on equal footing. the energy release rate G according to Eq. (S8). Then we The square root singularity is associated with a nite applied the tting procedure to the whole rupture prop- energy ux into the edge region, the so-called energy agation history and the a priori known value of G in release rate G, which for mode-III symmetry takes the Eq. (S2) has been used to plot in Fig. S2b G=G as a form [S8] function of L=L , where L is the rupture length. The results strongly support the expected relation G=G = 1 1 K III G(t) = : (S8) and hence also validate our tting procedure. Note that some deviation from G=G = 1 is observed, re ecting Our goal now is to extract the stress intensity factor from some uncertainly in the singular behavior, even in simple the singular elds of Eqs. (S6)-(S7), to use Eq. (S8) to slip-weakening models. Finally, for completeness, we also calculate G and to check whether the near-edge energy plot in Fig. S2b E (t)=c (t) of Eq. (S5), normalized by diss r balance G =G is satis ed. As all of the assumptions of G , which indeed equals unity throughout the rupture c c conventional fracture theory are satis ed by the model, propagation process, as expected. The same tting pro- the energy balance equation should be satis ed. cedure is applied in the manuscript to the frictional rup- We start by estimating the stress intensity factor from ture dynamics of interfaces described by rate-and-state the near-edge stress and slip velocity distributions shown friction, as discussed next. S3 1.4 0.8 -1 0.6 -4 -3 -2 10 10 10 0.4 0.6 0.2 0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 2 4 6 8 10 12 FIG. S2. (a) A snapshot of the normalized stress and slip velocity elds (see legend and the left-hand-sides of Eqs. (S6)-(S7)) near the edge of a rupture propagating at a speed c to the left (the snapshot corresponds to the blue horizontal line in Fig. S1, where rupture propagation in the simple slip-weakening cohesive zone model is presented). Note that  is used to nondimensionalize the elds and that  = 0 in this case. The black dashed lines correspond to ts to Eqs. (S6)-(S7), see text res for additional details. (inset) The same as the main panel, but on a double logarithmic scale and the x-axis is jx x j=W , see text for additional details. (b) G and E =c , both normalized by G , are plotted as a function of the normalized rupture size diss r c L=L (see legend in order to distinguish the di erent curves). These quantities are discussed in detail in the text. S-2. APPLICATION TO THE FRICTIONAL th length of Eq. (S3) takes the form RUPTURE DYNAMICS OF INTERFACES 4 G DESCRIBED BY RATE-AND-STATE FRICTION L = ; (S9) ( ) which is identical to the corresponding expression in A procedure similar to the one described in the pre- Eq. (6) in the manuscript, up to the dimensionless and vious section is applied in the manuscript to the fric- order unity pre-factor 4=. tional rupture dynamics of interfaces described by rate- As discussed in the manuscript, the generalized and-state friction. However, the di erences between the Grith-like length in Eq. (S9) and in Eq. (6) in the simple slip-weakening cohesive zone model discussed in manuscript highlights another di erence between simple the previous section and the more realistic rate-and-state slip-weakening cohesive zone models and rate-and-state friction models discussed in the manuscript, which are friction models related to G . While in slip-weakening intimately related to the central question addressed in cohesive zone models G is an a priori prescribed quan- the manuscript, call for some modi cations that will be tity, in rate-and-state friction models the existence and discussed here. First, frictional rupture features a nite identi cation of a well-de ned G from the interfacial dy- residual stress  > 0 under some conditions (extensively namics is not obvious. That is, one should understand res discussed in [S1]). That is, the strength of the inter- whether and how an e ective fracture energy G can be face does not drop to zero behind the rupture front as in properly de ned, and what the associated lengthscale is. the simple slip-weakening cohesive zone model (note that A procedure to de ne and extract G is discussed and in general slip-weakening cohesive zone models can de - employed in the manuscript. Here we supplement it with nitely feature a constant residual stress  ), but rather additional rationalization and details. res attains a nite value (on what lengthscale this value is The basic idea is related to the observation that the attained is yet another central question addressed in the frictional stress  (x; t) follows two distinct relaxation manuscript). The linearity of the elastodynamic eld regimes in the wake of rupture fronts, as demonstrated equations [S10] implies that the driving stress  in the in Fig. 3a in the manuscript. It rst undergoes a rather ordinary fracture case should be simply replaced by the strong initial drop that is followed by a slow decay to- stress drop  =  > 0 in the frictional case. This wards  . Such behavior is inherent to the rate-and- d res res implies that  should be subtracted from the stress state dependence of the frictional strength [S11]. The res eld  (x; t) before tting it to the square root singular initial strong drop is associated with a rather localized contribution in Eq. (S6) (cf. Fig. 3a in the manuscript). region near the rupture edge (see arrow in Fig. 3a in the Moreover, this implies that a generalization of the Grif- manuscript) and the slow decay towards  is charac- res S4 0.25 1.2 ing to v(x; t)(x; t)=D > 1, as formulated in Eq. (10) in the manuscript. 1.15 0.2 The latter criterion is demonstrated in Fig. S3, where 1.1 the frictional stress  (x; t) of Fig. 3a in the manuscript 0.15 -0.01 0 0.01 is superimposed on v(x; t)(x; t)=D, to exactly corre- 1.05 spond to the change in the relaxation behavior of  (x; t) 0.1 towards  that was discussed above. This criterion res is also in line with recent physics-based interpretations 0.05 of rate-and-state friction formulations [S12{S14]. Fi- 0.95 nally, for completeness, we present in Fig. S4 a snap- 0 0.9 shot of the spatial distribution of the real contact area -0.05 -0.04 -0.03 -0.02 -0.01 0 0.01 0.02 0.03 0.04 0.05 A (x; t) 1 + b log[1 + (x; t)= ] [S11]. 1.1 1.2 FIG. S3. A snapshot of the properly normalized (see legend) 1.6 stress eld  (x; t) (left y-axis) and v(x; t)(x; t)=D (right y- 1.15 1.4 1.08 axis) corresponding to the solution presented in Fig. 3a in the 1.2 manuscript, where the y-axis is truncated to allow the prop- 1.1 erties of the elds near the rupture edge to be visible. (inset) 1.06 -0.2 -0.1 0 0.1 v(x; t)(x; t)=D near the rupture edge without truncating the 1.05 y-axis. 1.04 1.02 0.95 terized by a much larger lengthscale. We consequently proposed that the former should be associated with the 0.9 e ective fracture energy G . -0.05 -0.025 0 0.025 0.05 In order to formalize this idea and to make the extrac- tion of G quantitative, we focus on the dimensionless FIG. S4. A snapshot of the real contact area A (x; t) 1 + combination v(x; t)(x; t)=D, which is shown in Fig. S3 b log[1 + (x; t)= ] (blue line, left y-axis) corresponding to and which according to Eq. (8) in the manuscript con- v(x; t)(x; t)=D of Fig. S3, which is reproduced here (orange trols the evolution of the structural state of the inter- line, right y-axis). The real contact area also exhibits slow face (x; t). The latter is known to determine the real relaxation to its asymptotic value behind the rupture edge. contact area A (x; t) 1 + b log[1 + (x; t)= ] of the in- (inset) A full scale plot of A (x; t) 1 + b log[1 + (x; t)= ] terface [S12] (for the de nition of the parameters b and near the rupture edge, directly demonstrating that the latter , and their values used here, see [S1, S2]). Hence, it is is associated with a reduction of the real contact area. directly related to the rupture process, involving a tran- sition from an initial value of A ahead of the rupture We note that the estimation of G through the dissi- front to a signi cantly lower value behind it (see the in- pation corresponding to the criterion v(x; t)(x; t)=D> 1 set of Fig. S4). This transition corresponds to a transi- appears to be consistent with available analytic approx- tion between v=D = 1 ahead of the rupture front, with imations for the e ective fracture energy [S15{S17]. In a very small v and hence a large , and v=D = 1 be- particular, the expression hind it, with a large v and hence a much smaller . In between, v=D is expected to attain signi cantly larger D @f (jvj; ) G = [log(v =v )] (S10) c c bg values. This physical picture is demonstrated in the inset 2 @ log () of Fig. S3, which corresponds to the rupture front shown in Fig. 3a in the manuscript. has been proposed in [S17]. Here @f (jvj; )=@ log () is The two-step nature of the approach of v=D to its the aging coecient (f (jvj; ) is the friction law intro- steady-state is revealed in the main panel of Fig. S3, duced in Eq. (7) in the manuscript), v corresponds to bg which presents a zoomed in version of the inset. The the steady-state velocity in the stick state (prior to the gure reveals that after the huge peak in v=D, which oc- arrival of the rupture front) and v is the slip velocity curs on a small lengthscale near the rupture edge, v=D far behind the rupture front. We estimate v as the bg undershoots unity and then approaches unity slowly from leftmost intersection point in Fig. 1b in the manuscript, below, on a signi cantly larger lengthscale. We conse- i.e. v  10 m/s, and v as the rightmost intersec- bg c quently attribute the small lengthscale weakening pro- tion point with the e ective steady-state friction curve, cess to the near-edge dissipation G , i.e. to the e ec- i.e. v  10 m/s. Using the parameters used in this c c 6 6 tive fracture energy, where the additional dissipation as- work (see [S2]), i.e. D = 0:5 10 m,  = 10 Pa and sociated with the larger lengthscale is discussed in the @f (jvj; )=@ log () = 0:021 (the latter equals bf in the manuscript. In quantitative terms, this picture implies notation of [S2]), and plugging everything in Eq. (S10), that G is estimated through the dissipation correspond- we obtain G  0:7J/m . The latter is in reasonably good c c S5 agreement with G of Fig. 3b in the manuscript. In or- tribution of near-edge elds and to test the energy bal- der to further substantiate this agreement, future work ance relation G =G presented in Sect. S-1 is applied in should extend the comparison by systematically varying the manuscript to rate-and-state frictional interfaces. In the parameters involved. this case,  is replaced by the stress drop  and G d c is estimated from the interfacial dynamics according to To conclude, the procedure to extract the singular con- Eq. (10) in the manuscript, as explained in detail here. [S1] F. Barras, M. Aldam, T. Roch, E. A. Brener, E. Bouch- [S10] A. C. Palmer and J. R. Rice, The Growth of Slip Sur- binder, and J.-F. Molinari, The emergence of crack- faces in the Progressive Failure of Over-Consolidated like behavior of frictional rupture: The origin of stress Clay, Proc. R. Soc. A Math. Phys. Eng. Sci. 332, 527 drops, To appear in Physical Review X (2019). (1973). [S2] E. A. Brener, M. Aldam, F. Barras, J.-F. Moli- [S11] T. Baumberger and C. Caroli, Solid friction from stick- nari, and E. Bouchbinder, Unstable Slip Pulses and slip down to pinning and aging, Adv. Phys. 55, 279 Earthquake Nucleation as a Nonequilibrium First-Order (2006). Phase Transition, Phys. Rev. Lett. 121, 234302 (2018). [S12] T. Baumberger and P. Berthoud, Physical analysis of [S3] D. Dugdale, Yielding of steel sheets containing slits, J. the state- and rate-dependent friction law. II. Dynamic Mech. Phys. Solids 8, 100 (1960). friction, Phys. Rev. B 60, 3928 (1999). [S4] G. Barenblatt, The Mathematical Theory of Equilib- [S13] Y. Bar-Sinai, R. Spatschek, E. A. Brener, and E. Bouch- rium Cracks in Brittle Fracture, Adv. Appl. Mech. 7, binder, On the velocity-strengthening behavior of dry 55 (1962). friction, J. Geophys. Res. Solid Earth 119, 1738 (2014). [S5] M. S. Breitenfeld and P. H. Geubelle, Numerical anal- [S14] A. Molinari and H. Perfettini, Fundamental aspects of ysis of dynamic debonding under 2D in-plane and 3D a new micromechanical model of rate and state friction, loading, Int. J. Fract. 93, 13 (1998). J. Mech. Phys. Solids 124, 63 (2019). [S6] F. Barras, D. S. Kammer, P. H. Geubelle, and J.-F. [S15] M. Cocco and A. Bizzarri, On the slip-weakening be- Molinari, A study of frictional contact in dynamic frac- havior of rate- and state dependent constitutive laws, ture along bimaterial interfaces, Int. J. Fract. 189, 149 Geophys. Res. Lett. 29, 1516 (2002). (2014). [S16] A. Bizzarri and M. Cocco, Slip-weakening behavior dur- [S7] D. J. Andrews, Rupture propagation with nite stress ing the propagation of dynamic ruptures obeying rate- in antiplane strain, J. Geophys. Res. 81, 3575 (1976). and state-dependent friction laws, J. Geophys. Res. [S8] L. B. Freund, Dynamic Fracture Mechanics (Cambridge 108, 2373 (2003). university press, Cambridge, 1998). [S17] A. M. Rubin and J.-P. Ampuero, Earthquake nucleation [S9] E. Jones, T. Oliphant, and P. Peterson, SciPy: Open on (aging) rate and state faults, J. Geophys. Res. Solid source scienti c tools for Python (2001). Earth 110, B11312 (2005).

Journal

Condensed MatterarXiv (Cornell University)

Published: Jul 9, 2019

References