Int J Fract (2010) 162:51–67 DOI 10.1007/s10704-009-9388-6
ORIGINAL PAPER
Slip modes and partitioning of energy during dynamic frictional sliding between identical elastic–viscoplastic solids Zheqiang Shi · Alan Needleman · Yehuda Ben-Zion
Received: 31 March 2009 / Accepted: 4 August 2009 / Published online: 10 September 2009 © Springer Science+Business Media B.V. 2009
Abstract The effect of plasticity on dynamic frictional sliding along an interface between two identical elastic–viscoplastic solids is analyzed. The configuration considered is the same as that in Coker et al. (J Mech Phys Solids 53:884–992, 2005) except that here plane strain analyses are carried out and bulk material plasticity is accounted for. The specimens have an initial compressive stress and are subject to shear loading imposed by edge impact near the interface. The material on each side of the interface is modeled as an isotropically hardening elastic–viscoplastic solid. The interface is characterized as having an elastic response together with a rate- and state-dependent frictional law for its inelastic response. Depending on bulk material properties, interface properties and loading conditions, frictional slip along the interface can propagate in a crack-like mode, a pulse-like mode or a train-of-pulses mode. Results are presented for the effect of material Z. Shi · Y. Ben-Zion Department of Earth Sciences, University of Southern California, Los Angeles, CA 90089, USA Z. Shi e-mail:
[email protected] Y. Ben-Zion e-mail:
[email protected] A. Needleman (B) Department of Materials Science and Engineering, College of Engineering and Center for Advanced Scientific Computing and Modeling, University of North Texas, Denton, TX 76203, USA e-mail:
[email protected];
[email protected]
plasticity on the mode and speed of frictional slip propagation as well as for the partitioning of energy components between stored elastic energy, kinetic energy, plastic dissipation in the bulk and frictional dissipation along the interface. Some parameter studies are carried out to explore the effects of varying the interface elastic stiffness and the impact velocity. Keywords Friction · Plasticity · Dynamical slip · Finite element modeling
1 Introduction Frictional sliding along an interface between two rapidly deforming solids is a basic problem in mechanics encountered in a variety of contexts and at a variety of scales. Rapid frictional sliding occurs on earthquake faults during propagation of earthquake ruptures between two opposing fault blocks loaded by tectonic stresses. Complex frictional sliding phenomena are observed over a wide range of size scales (see e.g. Persson 2000). Previous experimental and numerical studies, with a main focus on geophysical applications, have shown that frictional sliding on an interface between homogeneous elastic solids can occur in various modes including a crack-like mode, a pulse-like mode and a train-of-pulses mode, and also can propagate at various speeds including subshear and supershear. In a cracklike mode the slipping region continuously expands
123
52
along the interface and the slip duration at a given point in the sliding region is comparable to the overall slip duration. On the other hand, in a pulse-like slip mode [first proposed by Heaton (1990)] only a small portion of the interface slides at a given time and the slip duration at a given point in the sliding region is much smaller than the overall sliding duration. Factors controlling the selection of sliding mode and propagation speed include: (i) the interface constitutive properties, with a relatively strong dependence of friction on sliding speed favoring a pulse-like mode and a relatively weak dependence favoring a crack-like mode (e.g. Beeler and Tullis 1996; Zheng and Rice 1998; Nielsen et al. 2000; Ampuero and Ben-Zion 2008); (ii) relatively high and smooth stress loading favors supershear propagation while relatively low stress loading favors subshear propagation (e.g. Burridge 1973; Andrews 1976; Day 1982; Rosakis 2002; Xia et al. 2004; Shi et al. 2008); and (iii) the strength, abruptness and nature of the nucleation procedure used in earthquake modeling can have a strong effect on both the sliding mode and the speed of propagation (e.g. Shi and Ben-Zion 2006; Festa and Vilotte 2006; Shi et al. 2008). A contrast of elastic properties across the fault can also affect the sliding mode and the propagation speed (e.g. Weertman 1980; Ben-Zion 2001; Brietzke et al. 2009). The analyses of frictional sliding mentioned above assume that the bulk material is elastic. However, the elastically predicted stress concentration near the sliding interface may induce the bulk material to respond inelastically (Poliakov et al. 2002; Rice et al. 2005). The generation of inelastic strain may affect the sliding mode and the speed of propagation as well as the partitioning of energy. Several numerical studies have been carried out in the geophysics literature on the effects of off-fault inelastic response during crack-like dynamic ruptures of earthquake faults. These studies focused on the damage pattern in the bulk and treated off-fault inelastic response in various forms including formation of tensile cracks (e.g. Yamashita 2000; Dalguer et al. 2003), shear branches (e.g. Ando and Yamashita 2007) or generation of plastic strain using a Coulomb yielding criterion (e.g. Andrews 2005; Templeton and Rice 2008; Viesca et al. 2008; Duan and Day 2008). Offfault Coulomb yielding was also used in dynamic simulations of a pulse-like rupture on a frictional interface between two solids with different properties (Ben-Zion and Shi 2005).
123
Z. Shi et al.
Here, we carry out dynamic analyses of mode-II frictional sliding along an interface between two identical solids under plane strain conditions initiated by impact loading. The calculations employ the same configuration analyzed in Coker et al. (2005) but here plane strain conditions are considered. Two identical rectangular regions are held together by an initial compressive stress. Shear loading is imposed by edge impact near the interface. Loading conditions and parameter values are considered that, for purely elastic behavior, give rise to frictional slip in a crack-like mode, in a pulselike mode and in a train-of-pulses mode. The materials are modeled as isotropically hardening and rate dependent. Friction is characterized by a rate- and statedependent relation (Dieterich 1979; Ruina 1983; Rice and Ruina 1983; Linker and Dieterich 1992; Povirk and Needleman 1993), including the Prakash–Clifton (Prakash and Clifton 1993; Prakash 1998) normal stress dependence. Our investigation focuses on the effects of plasticity on the mode and speed of frictional sliding along the interface and on the partitioning of energy. We also carry out parameter studies exploring the effects of variations in interface elastic stiffness and impact velocity.
2 Problem formulation 2.1 Boundary value problem formulation A small strain formulation is used. In Cartesian tensor notation, the principle of virtual work is written as σi j δi j dv − Ti δu i dS = Ti δu i dS v
Sint
∂ 2ui − ρ 2 δu i dv ∂t
Sext
(1)
v
where t is time, ρ is the material density, σi j are the stress tensor components, u i are the displacement vector components, i j are the components of the infinitesimal strain tensor, denotes a jump across an interface, = ( )+ − ( )− , and v, Sext and Sint are the volume, external surface area and internal surface area, respectively. The density of the material is ρ and Ti = σi j n j with n j being the surface normal components. Computations are carried out for the specimen geometry shown in Fig. 1 with = 75 mm and w = 132 mm. Plane strain conditions are assumed and
Slip modes and partitioning of energy
53
Σ0
Fig. 1 Geometry of the configuration analyzed and time function of the impact loading. The dashed line denotes the interface Sint where frictional sliding occurs
x2 Sint
2l
b V(t)
V(t)
Vimp
tr
tp
ts
a Cartesian coordinate system is used as reference, with the x1 − x2 plane as the plane of deformation. For t ≤ 0, the body has a uniform compressive stress of magnitude 0 and this stressed state is used as a reference for deformation measurements. A normal velocity V (t) is prescribed on the edge x1 = 0, along the interval −b ≤ x2 ≤ 0, with b = 25 mm, where the shear traction vanishes, so that t u 1 = − V (ξ )dξ , T2 = 0 on x1 = 0 and 0
−b ≤ x2 ≤ 0.
x1
(2)
In Eq. (2) V (t) = ⎧ Vimp t/tr , for 0 ≤ t < tr , ⎪ ⎪ ⎪ ⎪ for tr ≤ t ≤ (t p + tr ), ⎨ Vimp , for (t p + tr ) < t < Vimp [1− (3) ⎪ ⎪ ⎪ − t )]/t , (t + t + t ), (t − t p r s r p s ⎪ ⎩ 0, for t ≥ (tr + t p + ts ), where, tr = 10µs is the rise time, t p = 50µs is the pulse width and ts = 10µs is the step-down time. On the remaining external surfaces, the tractions vanish, i.e. Ti = 0. The finite element discretization is based on linear displacement triangular elements that are arranged in a ‘crossed-triangle’ quadrilateral pattern. Along the interface (x2 = 0), four-node elements are used to give a linear dependence of the displacement jump on x1 . The mesh employed is the same as in Coker et al. (2005) which consists of a fine uniform region near the interface close to the side of impact connected to a coarse outer region by a transitional region with gradually varying mesh spacing. The domain analyzed has 18,320
w
t
Σ0
quadrilateral elements and 74,360 degrees of freedom. The uniform fine-mesh region is 60 mm × 4 mm and is discretized into 4,000 quadrilaterals, each of which is 0.3 mm × 0.2 mm. The bulk finite elements elements are constant strain elements and so, effectively, onepoint integration is used to form the element stiffness matrices. The interface element stiffness matrices are integrated using four-point Gaussian quadrature. The equations that result from substituting the finite element discretization into Eq. (1) are of the form ∂ 2U = R (4) ∂t 2 where M is a mass matrix, U is the nodal displacement vector and R is the nodal force vector. The equations of motion (4) are integrated numerically by an explicit integration procedure, the Newmark β-method with β = 0 (Belytschko et al. 1976). A lumped mass matrix is used instead of the consistent mass matrix, since this has been found preferable for explicit time integration procedures, from the point of view of accuracy as well as computational efficiency (Krieg and Key 1973).
M
2.2 Bulk elastic–viscoplastic constitutive relation The bulk material is characterized as an elastic– viscoplastic isotropically hardening solid. The total strain rate, ˙i j , is written as the sum of an elastic part, p ˙iej , and a plastic part, ˙i j , p
˙i j = ˙iej + ˙i j
(5)
with (˙) denoting partial differentiation with respect to time and 1+ν ν 3 ˙¯ p σ˙ i j − δi j σ˙ kk , ˙i j = ˙iej = si j . (6) E E 2 σe
123
54
Z. Shi et al.
In Eq. (6), δi j is the Kronecker delta, E is Young’s modulus, ν is Poisson’s ratio, ˙¯ is the effective plastic strain rate and σe is the Mises effective stress defined as 3 ˙¯ = 2 ˙ p ˙ p , σe = (7) si j si j ij ij 3 2 where si j is the deviatoric stress given by 1 si j = σi j − σkk δi j (8) 3 and ¯˙ = ˙0 [σ¯ /g(¯ )]1/m , g(¯ ) = σ0 (¯ /0 + 1) N . (9) In Eq. (9), σ0 and 0 are the reference stress and strain related by σ0 = E0 , and ¯ = ˙¯ dt is the effective plastic strain. The function g(¯ ) specifies the strain hardening characteristics with σe = g(¯ ) representing the effective stress versus effective strain response in a tensile test carried out at ˙¯ equal to the reference strain rate ˙0 , N is the strain hardening exponent and m is the strain-rate hardening exponent with m → 0 corresponding to the rate-independent limit. A rate tangent modulus method is used for the viscoplastic stress update (Peirce et al. 1984). Since the material response is viscoplastic, the speeds of dilational, shear and Rayleigh waves are still given by the elastic expressions (see e.g. Freund 1998) E(1 − ν) E , cs = , cl = ρ(1 + ν)(1 − 2ν) 2ρ(1 + ν) 0.862 + 1.14ν c R = cs . (10) 1+ν As in Coker et al. (2005), the elasticity parameters are chosen to represent Homalite-100: E = 5.3 GPa, ν = 0.35 and ρ = 1246 kg/m3 . The corresponding wave speeds given by Eq. (10) are cl = 2,613, cs = 1,255 and c R = 1,174 m/s. 2.3 Interface constitutive relations The interface Sint along x2 = 0 is characterized by an elastic and a rate- and state-dependent friction relation (Dieterich 1979; Ruina 1983; Rice and Ruina 1983; Linker and Dieterich 1992; Povirk and Needleman 1993). As in Povirk and Needleman (1993), Coker et al. (2005) and Shi et al. (2008), the traction rates along the interface are specified by (11) T˙n = −Cn u˙ n T˙s = −Cs [u˙ s − sgn(Ts )u˙ slip ].
123
(12)
Here, the subscripts s and n denote quantities tangential and normal to the interface, respectively, and Cs and Cn are the interface elastic stiffness parameters. Eqs. (11) and (12) apply if the surfaces are in contact; if they separate then Tn = Ts = 0. The jump of the tangential particle velocity across the interface, u˙ s , consists of an elastic part u˙ es and a frictional slip part u˙ slip such that u˙ s = u˙ es + u˙ slip ,
(13)
where u˙ es = −T˙s /Cs . We use the phenomenological expression for u˙ slip proposed by Povirk and Needleman (1993), and used in Coker et al. (2005), with a Prakash–Clifton (Prakash and Clifton 1993; Prakash 1998) regularization for the normal stress dependence. In Eq. (12), u˙ slip = 0 if Tn ≤ 0, while for Tn > 0, which corresponds to the interface being under compression, u˙ slip is given by
r
|Ts | u˙ slip = V0 −1 θPC g(θ0 ) |Ts | >1 (14) for θPC g(θ0 ) and u˙ slip = 0 otherwise. In Eq. (14) p 0 μd + (μs − μd ) exp − L 0V/θ 1 , (15) g(θ0 ) = 1/r L 0 /θ0 + 1 V0 where the evolution of the state variable θ0 is given by θ u˙ ˙θ0 = B 1 − 0 slip (16) L0 and the evolution of the Prakash–Clifton internal variable θPC is given by 1 (17) (θPC − Tn ) u˙ slip . L PC In Eqs. (12–17) μs and μd are the nominal static and dynamic coefficients of friction, L 0 is the characteristic slip distance over which the friction coefficient evolves to a steady-state value, L PC is the characteristic slip distance over which the normal stress evolves to a steady-state value, V0 and V1 are reference slip velocities. All the calculations in this study employ the parameter values μs = 0.6, μd = 0.5, V0 = 100 m/s, V1 = 26 m/s, L 0 = L PC = 0.2 mm, r = 5, p = 1.2 and B = 4.6. Unless otherwise specified, Cn = 300 GPa/m, Cs = 100 GPa/m. These parameter values were used in Coker et al. (2005) and a subset of these values (only a subset were relevant to the experiments) θ˙PC = −
Slip modes and partitioning of energy
55
were found to characterize experimental observations of Homalite sliding on Homalite (e.g. Rosakis et al. 2000; Rosakis 2002; Samudrala et al. 2002). Also note that a restricted form of the Prakash–Clifton (Prakash and Clifton 1993; Prakash 1998) response with only one state variable is employed. The interface shear traction Ts is updated using a rate tangent method the formulas for which are given in the Appendix. 2.4 Energy Partitioning One focus of this study is the effect of plasticity on energy partitioning which is of particular interest in earthquake dynamics (e.g. Husseini 1977; Shi et al. 2008). Following Shi et al. (2008), the energy balance equation can be written as Wboun −Welas − Wkin − Wplas + φelas + φslip = 0 (18) where the various energy components are defined as: – Energy input at the external boundaries ⎡ ⎤ t ⎢ ⎥ Wboun = ⎣ Ti u˙ i dS ⎦ dt 0
(19)
Sext
– Change of elastic strain energy and plastic dissipation in the bulk material ⎤ ⎡ t Welas = ⎣ σi j ˙iej dV ⎦ dt, 0
t Wplas =
V
⎤ ⎡ ⎣ σi j ˙ p dV ⎦ dt ij
0
(20)
V
– Kinetic energy (or radiated energy) 1 ρ u˙ i u˙ i dV Wkin = 2
(21)
V
– Change of interface energy consisting of a frictional part and an elastic part ⎡ ⎤ t ⎢ ⎥ (22) φslip = ⎣ Ts u˙ slip dS ⎦ dt 0
t φelas = 0
Sint
⎡ ⎢ ⎣
Sint
Ts u˙ s − sgn(Ts )u˙ slip dS
+
⎤ ⎥ Tn u˙ n dS ⎦ dt
(23)
Sint
3 Numerical results Calculations were carried out for three combinations of initial compressive stress 0 and impact velocity Vimp considered in Coker et al. (2005) Case I: A crack-like mode— 0 = 6 MPa, e = 5.27 MPa, Vimp = 2 m/s. Case II: A pulse-like mode— 0 = 30 MPa, e = 26.4 MPa, Vimp = 2 m/s. Case III: A train-of-pulses mode— 0 = 10 MPa, e = 8.79 MPa, Vimp = 20 m/s. where e is calculated from 0 using Eq. (7). Our aim is to explore the effect of bulk material inelasticity on the mode of frictional sliding and on the partitioning of energy rather than to describe the response of a particular material. For each case we present results for elastic material behavior and results where the bulk material reference strength σ0 is taken to be slightly greater than e , the initial effective stress associated with the initial stress state. We employ a simple isotropic hardening material characterization and, since the initial stress state differs in each case, the material description is taken to vary from case to case. The term used to characterize each case is the frictional slip mode obtained in the elastic calculations. The strain hardening exponent N is fixed at 0.1, to give a moderately strain hardening response, and the reference strain rate is ˙0 = 1 s−1 . In Cases I and III, the strain-rate hardening exponent m is taken to be 0.01 giving a nearly rate-independent response, while in Case II, for numerical stability reasons, m is set to 0.05 giving a stronger rate dependence. The calculations are carried out using an initial time step of 0.05 × min(ds)/cl in Cases I and III and 0.002 × min(ds)/cl in Case II, where min(ds) denotes the shortest edge length of the triangular elements. An adaptive time step scheme is employed that limits the maximum value of u˙ slip dt in a single time step to 0.01L 0 . In all cases an elastic wave propagates into the bulk material and along the interface with the dilational wave speed cl at the initiation of impact. The loading wave induces an elastic displacement jump, u es , along the interface Sint and a corresponding interfacial shear
123
56
0.6 0.4 0.2 12 10 8 6 4 2 0
3.1 Case I: a crack-like mode
123
0 20 μs 25 μs 30 μs
10
20
30
40
50
60
70
x1(mm)
0.6 0.4 0.2 12 10 8 6 4 2 0
Ts / Tn
(b)
Δuslip (m/s)
With 0 = 6 MPa and Vimp = 2 m/s, a crack-like slip mode is obtained in both the elastic calculation and the calculation with σ0 = 5.3 MPa (slightly above e ). Figure 2 plots the frictional slip rate, u˙ slip , along the interface together with the apparent coefficient of friction, μapp = Ts /Tn , at different times, for both elastic and plastic calculations. The slip rate with σ0 = 5.3 MPa is greater than in the elastic calculation and there is a pronounced peak in slip rate near the frictional slip front. In addition, with σ0 = 5.3 MPa there is a small region along Sint near the edge of impact where u˙ slip = 0 due to a loss of contact so that Tn = Ts = 0 and μapp is undefined. In the elastic calculation, frictional sliding nucleates from the boundary near the impact edge (x1 = 0) at t = 15.72 µs when Ts increases to 3.6 MPa and Tn decreases to 5.7 MPa so that μapp = 0.63 there. With σ0 = 5.3 MPa frictional sliding nucleates at t = 12.39 µs, earlier than in the elastic calculation, with Tn = Ts = 3.6 MPa (μapp ≈ 1) at the impact end of the interface. In both calculations, u˙ slip increases sharply after nucleation while u˙ es decreases sharply to slightly below zero and remains small over most of the region with frictional sliding. Both Ts and Tn in the calculation with σ0 = 5.3 MPa are smaller everywhere along Sint than in the elastic calculation and vanish near the impact end. Figure 3 shows curves of frictional slip propagation speeds Vslip for both calculations. The frictional slip propagation speed is calculated by first determining the position of a frictional slip front, defined as the furthest distance from the impact end where u˙ slip exceeds a specified threshold value. A seven-point progressive least-squares fit to the slip front position as a function of time is then used to calculate Vslip . In the calculations here, the threshold value of u˙ slip is chosen to be 5 × 10−4 m/s. Calculations using other threshold
0
Ts / Tn
(a)
Δuslip (m/s)
traction Ts . After some time, which depends on the loading condition and on the properties of the bulk material and the interface, frictional sliding initiates and propagates along the pre-loaded interface. The calculations are terminated before the frictional slip front reaches the end of the uniform fine-mesh region at x1 = 60 mm. We calculate energy changes both in the bulk and on the interface from the state at t = 0.
Z. Shi et al.
0 20 μs 25 μs 30 μs
0
10
20
30
40
50
60
70
x1(mm) Fig. 2 Along-interface profiles of u˙ slip and Ts /Tn in Case I at 20, 25 and 30 μs for a the elastic calculation and b the calculation with σ0 = 5.3 MPa
values of u˙ slip show little difference in the obtained slip propagation speeds. This insensitivity to the choice of threshold value of u˙ slip is due to the sharpness of the frictional slip front as shown in Fig. 2. For the elastic calculation, the calculated frictional slip propagation speed initially significantly exceeds the elastic longitudinal wave speed but then decreases quickly as slip propagates and attains a quasi-steady value close to cl . On the other hand, the calculated frictional slip propagation speed with σ0 = 5.3 MPa is initially less than cl but increases to a value close to but slightly less than that of the elastic calculation. As noted in Coker et al. (2005), the initiation process of frictional sliding under impact loading can involve coalescence of small sliding regions. Therefore the calculated apparent propagation speed of frictional slip front during the early stages can exceed cl .
Slip modes and partitioning of energy
57
7
(a)
14
elastic σ0 = 5.3 MPa
5 4 3
Welas
φ slip
10
Energy (J/m)
Vslip (km/s)
Wplas
12
6
Wkin
8 6 4 2 0
cl
−2 0
2
5
10
15
20
25
30
35
40
30
35
40
Time (μs) 1 5
10
15
20
25
30
35
40
(b)
Fig. 3 Frictional slip propagation speed, Vslip , versus time in Case I for the elastic calculation and the calculation with σ0 = 5.3 MPa
Figure 4 shows time histories of various energy components including the change of elastic strain energy in the bulk Welas , the plastic dissipation in the bulk Wplas , the kinetic energy Wkin , and the frictional dissipation along the interface φslip , for both the elastic calculation and the calculation with σ0 = 5.3 MPa. In the elastic calculation (Fig. 4a), during the early stages of loading before the initiation of frictional slip, the energy input is divided between Welas and Wkin . Once frictional slip initiates, the energy dissipation from frictional sliding increases and eventually exceeds both Welas and Wkin . The energy balance involving all the energy terms in Eq. (18) was checked in this and all other calculations and the mismatch is less than 0.5% of the bulk energy (Welas + Wplas ). Figure 4b shows that in the calculation with σ0 = 5.3 MPa the bulk elastic energy initially decreases, due to plastic relaxation, and then increases somewhat, but the bulk energy is mainly plastic dissipation. Contours of Mises effective stress σe , defined in Eq. (7), are shown in Fig. 5a for the elastic calculation and in Fig. 5b for the calculation with σ0 = 5.3 MPa. Plasticity limits the effective stress levels that can be attained so that the scale in Fig. 5b is lower than that in Fig. 5a. In both calculations, the highest values of σe occur for x2 > 0 which is the half of the body above the impact surface. Stress relaxation occurs at the interface near the impact surface in Fig. 5b, which leads to the arrest of frictional sliding in a small region. Although contours of plastic strain rate and plastic strain for
Energy (J/m)
Time (μs)
18 16 14 12 10 8 6 4 2 0 −2
Wplas Welas
φ slip
Wkin
0
5
10
15
20
25
Time (μs) Fig. 4 Time histories of various energy components in Case I for a the elastic calculation and b the calculation with σ0 = 5.3 MPa
Case I are not shown, the maximum plastic strain is about 0.01 and the maximum plastic strain rate is about 2 × 103 s−1 .
3.2 Case II: a pulse-like mode In Case II, the initial stress is increased to 0 = 30 MPa with the impact velocity remaining at Vimp = 2 m/s. For numerical stability reasons, the rate sensitivity parameter in the calculation with σ0 = 27 MPa is taken to be m = 0.05, thus increasing the material strain rate sensitivity. In the elastic calculation sliding occurs in a pulselike mode as shown in Fig. 6a. Instead of nucleating at the boundary as in Case I, frictional sliding in the elastic calculation initiates as a small frictional slip patch at t = 29.55 μs at about x1 = 5.7 mm when Ts reaches a peak value of 18 MPa there. The slip patch expands in both directions with an asymmetric sliprate profile in a shape that has a gentle slope away from
123
58
Z. Shi et al.
(a)
(a)
x2 (mm)
14 12 10 8 6 4 2
0
−20
0.4 0.2
Ts / Tn
σe (MPa)
20
Δuslip (m/s)
0.6 1600 1400 1200 1000 800 600 400 200 0
0 slip patch
0
10
20
30 μs 35 μs 40 μs
30
40
50
60
70
x1(mm)
(b)
−40 40
x1(mm)
60
80
Δuslip (m/s)
20
(b)
x2 (mm)
20
σe (MPa) 6 5 4 3 2 1
0
−20
−40 0
20
40
x1(mm)
60
80
Fig. 5 Contours of Mises effective stress in Case I for a the elastic calculation at t = 34.6 μs and b the calculation with σ0 = 5.3 MPa at t = 34.6 μs. The frictional sliding front is at x1 = 58.0 mm in a and at x1 = 55.5 mm in b
the loading edge and a steep slope towards the loading edge. The amplitude of the steep-slope end of the frictional slip patch increases giving the slip patch a pulselike shape as it approaches the impact edge while the gentle-slope end of the slip patch maintains its shape. At t = 30.73 μs, the pulse-like slip patch meets the boundary with a peak frictional slip rate of 1,325 m/s and its single peak starts to split into two peaks creating two pulses going in opposite directions: one exits the impact end of the boundary and the other propagates away from that boundary with a peak frictional slip rate of about 600 m/s. The shear traction (and the shear traction rate) has a local minimum at the location of the slip pulse. As the frictional slip pulse propagates, the local minimum of shear traction moves with the pulse and the amplitude of a second peak of shear traction that
123
1600 1400 1200 1000 800 600 400 200 0 0
0.4 0.2
Ts / Tn
0.6
0
0 30 μs 35 μs 40 μs
10
20
30
40
50
60
70
x1(mm)
Fig. 6 Along-interface profiles of u˙ slip and Ts /Tn in Case II at 30, 35 and 40 μs for a the elastic calculation and b the calculation with σ0 = 27 MPa
follows behind the pulse gradually increases to nucleate another frictional slip patch that leads to a second pulse. This process repeats to generate more pulses. The nucleation location of the slip patch moves away from the impact end: the second, third and fourth slip patches nucleate at about x1 = 6.5, 8.5 and 12.5 mm. A developing slip patch can be seen in the slip-rate profile at 30 μs in Fig. 6a. As shown in Fig. 6b, the pulse-like mode of propagation is preserved in the calculation with σ0 = 27 MPa. The initiation of frictional sliding in the calculation with σ0 = 27 MPa occurs in a similar fashion to that in the elastic calculation. A small frictional slip patch nucleates around x1 = 2.5 mm at t = 28.32 μs when Ts reaches a peak value of 18 MPa there. Since the first slip patch is closer to the impact edge, it reaches the boundary only 0.63 μs after nucleation at t = 28.95 μs, when the peak slip rate exceeds 10, 000 m/s. The second slip patch nucleates at about the same location as the first slip patch when the second peak of Ts reaches 16 MPa and μapp reaches 0.65 there. The third and ensuing slip patches, however, nucleate at the boundary due to the rapid decrease of Tn there which gives rise to a large value of μapp . The time interval between the nucleation of slip patches
Slip modes and partitioning of energy
59
10
(a)
9
Vslip (km/s)
Energy (J/m)
elastic σ0 = 27 MPa
8 7 6 5 4 3
cl
2 1 25
30
35
40
180 160 140 120 100 80 60 40 20 0 −20
Wplas Welas
φ slip
Wkin
0
45
5
10 15 20 25 30 35 40 45
Time (μs)
Time (μs)
Fig. 7 Frictional slip propagation speed, Vslip , versus time in Case II for the elastic calculation and the calculation with σ0 = 27 MPa
decreases leading to a smaller spacing between pulses. Since the initial and subsequent slip patches have less time to grow than in the elastic calculation (because they nucleate closer to the boundary), the slip pulse has a smaller width than in the elastic calculation. By the time the first frictional slip pulse reaches the end of the fine-mesh region, there are eight fully developed pulses and one developing pulse in the calculation with σ0 = 27 MPa. By way of contrast, in the corresponding elastic calculation there are four fully developed pulses and one developing pulse. For both the elastic calculation and the calculation with σ0 = 27 MPa the shear traction rate T˙s is negative at the location of the frictional slip pulses and therefore so is u˙ es . In fact, as also seen by Coker et al. (2005), the magnitude of u˙ es is such that the velocity jump across the interface, u˙ s , exhibits a rather smooth variation along the interface despite the pulse nature of u˙ slip . The numerical issue leading to the use of m = 0.05 is due to high pulse frictional slip rates and corresponding high bulk plastic strain rates which necessitate extremely small time steps for numerical stability (with the algorithms used here). The plastic strain rate increases with decreasing m. For example, with m = 0.05 as in the calculation here the peak plastic strain rate is about 104 s−1 , while with m = 0.01 the peak plastic strain rate can reach 108 s−1 . For both the elastic calculation and the calculation with σ0 = 27 MPa, the calculated frictional slip propagation speed significantly exceeds cl over the entire time interval calculated (Fig. 7). Figure 8 shows the rapid rise in the energy dissipated in slip, φslip , once frictional slip propagation initiates.
Energy (J/m)
(b) 180 160 140 120 100 80 60 40 20 0 −20 0
Wplas Welas
φ slip
Wkin
5
10 15 20 25 30 35 40 45
Time (μs) Fig. 8 Time histories of various energy components in Case II for a the elastic calculation and b the plastic calculation with σ0 = 27 MPa
As for the elastic calculation in Case I, the values of Welas and Wkin remain close over the entire range of the computation. The absolute values of the various energy contributions are not directly comparable to those in Case I because the initial stress states differ. Comparing the energy partitioning in Fig. 8 shows that in the calculation with σ0 = 27 MPa the bulk stress working is mainly plastic dissipation rather than stored elastic energy. The time evolution of kinetic energy is not much changed. Bulk material plasticity results in greater frictional dissipation along the interface at a given time but this is mainly a shift due to the earlier nucleation time of frictional slip in the calculation with σ0 = 27 MPa than in the elastic calculation. The distributions of Mises effective stress σe in Fig. 9 reflect the pulse nature of frictional slip propagation for both the elastic calculation and the calculation with σ0 = 27 MPa. As in Case I, the effective stress levels are somewhat reduced in the calculation with σ0 = 27 MPa and the high values of σe mainly occur in the region x2 > 0.
123
60
Z. Shi et al.
0.6 0.4
σe (MPa) 48 46 42 40 35 30 25
0
−20
0.2
400
Δuslip (m/s)
x2 (mm)
20
Ts / Tn
(a)
(a)
0 300
10 μs 20 μs 29.4 μs
200 100 0 0
10
20
30
40
50
60
70
x1(mm)
(b) 0
20
40
x1(mm)
60
0.6
80
0.4
38 36 34 32 30 28 26
0
−20
Δuslip (m/s)
σe (MPa)
20
x2 (mm)
0.2
400
(b)
Ts / Tn
−40
0 300
10 μs 20 μs 30 μs
200 100 0 0
10
20
30
40
50
60
70
x1(mm)
Fig. 10 Along-interface profiles of u˙ slip and Ts /Tn in Case III at a 10, 20 and 29.4 μs for the elastic calculation and b 10, 20 and 30 μs for the calculation with σ0 = 10 MPa
−40 0
20
40
x1(mm)
60
80
Fig. 9 Contours of Mises effective stress in Case II for a the elastic calculation at t = 37.6 μs and b the calculation with σ0 = 27 MPa at t = 37.8 μs. The frictional sliding front is at x1 = 42.2 mm in a and at x1 = 42.4 mm in b
3.3 Case III: a train-of-pulses mode In Case III, the initial compressive stress 0 is 10 MPa but the impact velocity is 20 m/s, an order of magnitude larger than in Cases I and II. In the elastic calculation, frictional sliding nucleates near the boundary at t = 7.06 μs and propagates in a crack-like fashion. The value of μapp at the boundary continues to increase until t = 11.42 μs when it reaches a peak value of 0.54. Then Ts abruptly decreases while Tn remains relatively constant leading to a sharp decrease of μapp to about 0.5, which results in u˙ slip vanishing and leaving the initial cracklike region detached from the following slip. At the boundary, Tn monotonically decreases while Ts oscillates with an overall decreasing amplitude, which leads
123
to μapp oscillating between 0.46 and 0.62 generating a train of pulses as shown in Fig. 10a. Unlike the situation in Case I and Case II, Tn remains above its initial value along the entire interface throughout the calculation. With σ0 = 10 MPa, frictional sliding initiates from the boundary earlier at t = 6.3 μs. Also the oscillation of Ts occurs at a much shorter interval after initiation with an amplitude not large enough to bring u˙ slip to zero. Therefore the frictional slip mode remains cracklike overall but with strong oscillations near the leading front. A region with vanishing Ts and Tn occurs near the impact end of the interface at t = 13.6 μs. As slip continues to propagate, the oscillations of Ts increase eventually transforming the oscillatory portion of the crack into a train of pulses as shown in Fig. 10b. The peak frictional slip rate in a pulse in both calculations is about half that for the pulse-like propagation in Case II (Sect. 3.2). In the calculation with σ0 = 10 MPa, the peak frictional slip rate at t = 30 μs in the crack-like slip region is a factor of 4–5 times smaller than in the region with a train of pulses, and the pulses near the leading frictional slip front are
Slip modes and partitioning of energy
61
4
(a)
Vslip (km/s)
3.5 3
cl
2.5
Energy (J/m)
elastic σ0 = 10 MPa
2 1.5 0
5
10
15
20
25
450 400 350 300 250 200 150 100 50 0
30
Wplas Welas
φ slip
Wkin
0
5
10
Time (μs)
more closely spaced than in the corresponding elastic calculation. Compared to Fig. 3 in Case I, the initial evolution of the frictional slip propagation speeds in Case III for both the elastic calculation and the calculation with σ0 = 10 MPa shown in Fig. 11 are similar. In Case III the propagation speeds are more oscillatory and slightly slower with mean values of about 2–4% below cl . For the elastic calculation shown in Fig. 12a, as in the previous elastic calculations, the bulk elastic energy Welas and the kinetic energy Wkin are almost equal. However, for the train-of-pulses mode, the frictional dissipation remains much less than either Wkin or Welas over the entire time range of the computation. For the calculation with σ0 = 10 MPa shown in Fig. 12b, the bulk elastic and kinetic energies dominate the energy distribution during the early stages of propagation (t ≤ 10 μs) but in the latter stages, e.g. t > 25 μs, when the crack-like region has become more prominent, the bulk plastic dissipation and frictional dissipation on the interface dominate. At t = 30 μs, the frictional dissipation is about a factor of two greater than it is for the corresponding elastic calculation and the kinetic energy is similarly reduced. The distribution of Mises effective stress in Fig. 13a for the elastic calculation reflects the train-of-pulse propagation mode of slip. In contrast, for the calculation with σ0 = 10 MPa, the oscillations that are indicative of pulses are smoothed out and the distribution resembles that of the crack-like mode in Fig. 5b. Larger plastic strains develop in Case III than in Case II (values of effective plastic strain ¯ > 0.10) and plastic straining
20
25
30
(b) Energy (J/m)
Fig. 11 Frictional slip propagation speed, Vslip , versus time in Case III for the elastic calculation and the calculation with σ0 = 10 MPa
15
Time (μs)
450 400 350 300 250 200 150 100 50 0
Wplas Welas
φ slip
Wkin
0
5
10
15
20
25
30
35
Time (μs) Fig. 12 Time histories of various energy components in Case III for a the elastic calculation and b the calculation with σ0 = 10 MPa
also occurs over a larger region which acts to smooth the oscillations.
3.4 Parameter studies 3.4.1 Effect of interface elasticity In order to explore the effect of interface elasticity, we repeat the calculations in Case III using various values for the interface stiffness parameters Cs and Cn . Figure 14 shows the slip-rate profiles from elastic calculations and from calculations with σ0 = 10 MPa using Cs = 100 (the value used in the calculations in Sect. 3.3), 200 and 400 GPa/m with all other parameters the same as in Sect. 3.3. For the elastic calculations in Fig. 14a, frictional slip propagates as a train of pulses with Cs = 100 GPa/m. When Cs is increased to 200 GPa/m, the frictional slip mode changes to a crack-like mode with notable oscillations in the sliprate profile around x1 = 11 mm. When Cs is increased
123
62
Z. Shi et al.
(a)
(a) σe (MPa) 42 41 40 35 30 25 20 15 10 5
0
−20
350
250 200 150 100 50 0
−40 0
20
40
x1(mm)
60
0
10
20
80
30
(b)
14 12 10 8 6 4 2
−20
60
200
Δuslip (m/s)
σe (MPa)
0
50
Cs = 100 GPa, t = 26.57 μs Cs = 200 GPa, t = 21.95 μs Cs = 400 GPa, t = 23.70 μs
250
20
40
x1(mm)
(b)
x2 (mm)
Cs = 100 GPa, t = 25.72 μs Cs = 200 GPa, t = 23.34 μs Cs = 400 GPa, t = 21.74 μs
300
Δuslip (m/s)
x2 (mm)
20
150 100 50 0 0
10
20
30
40
50
60
x1(mm)
−40 0
20
40
x1(mm)
60
80
Fig. 13 Contours of Mises effective stress in Case III for a the elastic calculation at t = 29.4 μs and b the calculation with σ0 = 10 MPa at t = 30 μs. The frictional sliding front is at x1 = 59.1 mm in a and at x1 = 58.2 mm in b
further to 400 GPa/m, a much smoother crack-like sliprate profile is obtained. Similarly, for the calculations with σ0 = 10 MPa in Fig. 14b, the train of pulses in the front part of the slipping region are smoothed out and frictional slip propagates in a fully crack-like mode with a non-slip region near the impact end. Figure 15 shows that frictional slip initiates earlier and propagates slightly faster with fewer oscillations for larger values of Cs . Increased values of Cs result in decreased values of u˙ es which means that u˙ slip → u˙ s as Cs is increased. Thus, the transition to a smoother crack-like profile of u˙ slip with increasing Cs is consistent with u˙ s tending to remain smooth as noted in Sect. 3.2. Although the mode of frictional slip is affected by varying Cs , at least for the range of values considered here, the partitioning of energy does not vary much.
123
Fig. 14 Along-interface profiles of u˙ slip with the frictional slip front position at 50 mm in the a elastic calculations and b calculations with σ0 = 10 MPa. Vimp = 20 m/s and Cs = 100, 200 and 400 GPa/m. The time for each profile is indicated
In contrast to the effect of parameter Cs , the dynamics of the frictional sliding are not sensitive to variations in the parameter Cn . We repeated the calculations for Case III using Cn = 600 and 1,200 GPa/m. The frictional slip propagation speed and the slip mode are essentially unaffected by values of Cn in this range. There are, however, some small differences in the amplitude and phase of each pulse in the pulse train with different values of Cn . For the elastic calculations, the pulses have slightly larger peak rates with larger Cn . For the calculations with σ0 = 10 MPa, the pulse train near the rupture front is shorter with larger values of Cn . These small differences may be numerical artifacts since the adaptive time step algorithm results in smaller time steps for larger values of Cn . 3.4.2 Effect of impact velocity To explore the effect of impact velocity, the elastic calculation and the calculation with σ0 = 10 MPa
Slip modes and partitioning of energy
Vslip (km/s)
3.5 3
cl
2.5
(a) 0.6 0.4 0.2 60
Ts / Tn
Cs = 100 GPa/m, elastic Cs = 100 GPa/m, σ0 = 10 MPa Cs = 200 GPa/m, elastic Cs = 200 GPa/m, σ0 = 10 MPa Cs = 400 GPa/m, elastic Cs = 400 GPa/m, σ0 = 10 MPa
Δuslip (m/s)
4
63
0 10 μs 20 μs 30 μs
40 20 0 0
2
10
20
30
40
50
60
70
x1(mm)
(b) 0
5
10
15
20
25
0.6
30
0.4
Time (μs)
0.2 60
Δuslip (m/s)
Fig. 15 Frictional slip propagation speed, Vslip , versus time for elastic calculations and calculations with σ0 = 10 MPa with varying Cs
Ts / Tn
1.5
0 10 μs 20 μs 30 μs
40 20 0 0
10
20
30
40
50
60
70
x1(mm)
Fig. 16 Along-interface profiles of u˙ slip and Ts /Tn at 10,20 and 30 μs for a the elastic calculation and b the calculation with σ0 = 10 MPa. Vimp = 5 m/s but all other parameters the same as for Case III
(a) 0.4 0.2
Δuslip (m/s)
1000 800
10 μs 20 μs 28.5 μs
600 400
Ts / Tn
0.6
0
200 0 0
10
20
30
40
50
60
70
x1(mm)
(b) 0.4 0.2
1000 800
10 μs 20 μs 28.5 μs
600 400
Ts / Tn
0.6
Δuslip (m/s)
in Sect. 3.3 for Case III (where Vimp = 20 m/s) are repeated with Vimp = 5 and 30 m/s. In the elastic calculation frictional slip propagates in a crack-like mode with Vimp = 5 m/s (Fig. 16a), in contrast to propagating in a train-of-pulses mode with Vimp = 20 m/s (Fig. 10a) and 30 m/s (Fig. 17a). With Vimp = 30 m/s the pulses are more closely spaced and have higher peak frictional slip rates than in the corresponding calculation with Vimp = 20 m/s. With σ0 = 10 MPa and Vimp = 5 m/s, the profile of the slip rate becomes very oscillatory and the frictional slip mode begins to develop into a train of pulses as shown in Fig. 16b. When Vimp is increased to 30 m/s as shown in Fig. 17b, the slip-rate profile is much like that in Fig. 10b where Vimp = 20 m/s. Figure 18 shows that frictional slip initiates earlier with increasing impact velocity for both the elastic calculation and the calculation with σ0 = 10 MPa. In both cases, the slip propagation speeds with Vimp = 20 and 30 m/s, approach essentially the same value, which is slightly lower than the corresponding calculations with Vimp = 5 m/s. Thus, at least in the cases here, frictional slip in a crack-like mode propagates at a slightly greater speed than propagation in a train-of-pulses mode even though the impact velocity is smaller. As seen in Fig. 19 there is an increasing difference between the predictions of the elastic calculations and the calculations with σ0 = 10 MPa for the energy partitioning with increasing impact velocity: plastic dissipation increases, the kinetic energy of the calculation
0
200 0 0
10
20
30
40
50
60
70
x1(mm)
Fig. 17 Along-interface profiles of u˙ slip and Ts /Tn at 10,20 and 28.5 μs for a the elastic calculation and b the calculation with σ0 = 10 MPa. The impact velocity, Vimp , is 30 m/s and all other parameters are the same as in Sect. 3.3
123
64
Z. Shi et al.
Vslip (km/s)
3.5 3
cl
2.5
(a) 900
Vimp = 5 m/s, elastic Vimp = 5 m/s, σ0 = 10 MPa
800
Vimp = 20 m/s, elastic Vimp = 20 m/s, σ0 = 10 MPa Vimp = 30 m/s, elastic Vimp = 30 m/s, σ0 = 10 MPa
700 600
W kin (J/m)
Vimp = 5 m/s, elastic Vimp = 5 m/s, σ0 = 10 MPa Vimp = 20 m/s, elastic Vimp = 20 m/s, σ0 = 10 MPa Vimp = 30 m/s, elastic Vimp = 30 m/s, σ0 = 10 MPa
4
500 400 300 200 100
2
0 0
1.5
5
10
15
20
25
30
35
25
30
35
Time (μs)
(b) 0
5
10
15
20
25
30
35
Time (μs) Fig. 18 Frictional slip propagation speed, Vslip , versus time for the elastic calculation and the calculation with σ0 = 10 MPa with various values of Vimp
with σ0 = 10 MPa becomes increasingly less than that of the corresponding elastic calculation and the frictional dissipation of the calculation with σ0 = 10 MPa becomes increasingly greater than that of the corresponding elastic calculation.
1200
Vimp = 5 m/s, elastic Vimp = 5 m/s, σ0 = 10 MPa
1000
φslip (J/m)
1
Vimp = 20 m/s, elastic Vimp = 20 m/s, σ0 = 10 MPa Vimp = 30 m/s, elastic Vimp = 30 m/s, σ0 = 10 MPa
800 600 400 200 0 0
5
10
15
20
Time (μs)
Fig. 19 The effect of impact velocity Vimp on the time variation of a the kinetic energy Wkin and b the frictional dissipation along the interface φslip
4 Concluding remarks Two identical elastic–viscoplastic planar blocks having an initial compressive stress were considered. Plane strain conditions were assumed and frictional sliding along the interface was initiated by edge impact loading. Three sets of parameters were analyzed that, for elastic material behavior, gave rise to the same frictional slip modes as obtained in the plane stress calculations of Coker et al. (2005), i.e. a crack-like mode, a pulse-like mode and a train-of-pulses mode. In each case two sets of calculations were carried out, an elastic calculation and a plastic calculation with a reference flow strength slightly greater than the effective stress associated with the initial stress state. The effects of bulk material plasticity on the frictional slip mode and on the energy partitioning were investigated. Also, variations in interface elastic properties and in impact velocity were explored. With bulk plasticity, frictional slip initiates earlier but the slip propagation speeds are not much affected. In the plastic calculations, the values of both Ts and Tn along the interface are smaller than in the
123
corresponding elastic calculations, and when the frictional slip mode is crack-like a zone of vanishing tractions along the interface develops from the impact end of the interface. In Case I (crack-like propagation) and in Case II (pulse-like propagation), the impact velocity is relatively low, 2 m/s, and plastic deformation is rather limited (Fig. 20a). The frictional slip mode and the energy partitioning are not much affected by bulk plastic dissipation although the frictional slip rates are somewhat greater for plastic material behavior than for elastic material behavior. In Case III (train-of-pulses propagation), where the impact velocity is 20 m/s and plastic deformation is more extensive (Fig. 20b), there is a transition of the frictional slip mode from a train-ofpulses mode to a crack-like mode in the calculation with bulk plasticity. Also, the more extensive plastic deformation in this case results in an increase in frictional dissipation along the interface and a reduction in kinetic energy as compared with the corresponding elastic calculation.
Slip modes and partitioning of energy
65
When u˙ slip has an abrupt change, the elastic slip rate u˙ es changes in the opposite direction with a shape and amplitude similar to that of the frictional slip rate u˙ slip ; for example when a pulse develops or when u˙ slip abruptly vanishes near the impact end. A consequence of this is that the jump of the tangential velocity u˙ s maintains a relatively smooth profile and is cracklike along the interface. Calculations were carried out with the mesh size halved in each direction for a crack-like case and for a pulse-like case with elastic bulk material behavior. For crack-like propagation the results for the two meshes essentially coincided while for the pulse-like case there was a small difference in the pulse profile along the interface. Calculations for pulse-like frictional slip propagation with bulk material elasticity were also carried out with initial time step values varying from 0.05 × min(ds)/cl to 0.001 × min(ds)/cl where min(ds) is the minimum finite element edge length. The results agreed except for a small differences in the pulse amplitude in the early stages of slip. The effects of varying the interface elastic parameters and the impact velocity were explored. Figure 21 shows that large values of Cs tend to favor the crack-like mode of sliding. In the plastic calculations, increased impact velocity gives rise to increased bulk plastic deformation, increased frictional dissipation along the interface and a reduction in kinetic energy. The effect of Vimp on the frictional slip mode appears to be more complicated. However, based on results for elastic calculations with Cs = 200 GPa/m and plastic
(a)
x2 (mm)
20
ε 0.012 0.01 0.008 0.006 0.004 0.002
0
−20
−40 0
40
20
60
80
x1(mm) (b) ε
x2 (mm)
20
0.04 0.02 0.01 0.008 0.006 0.004 0.002
0
−20
−40 0
40
20
60
80
x1(mm) Fig. 20 Contours of effective plastic strain for a Case II with σ0 = 27 MPa at t = 37.8 μs and b Case III with σ0 = 10 MPa at t = 30 μs. In a the sliding front is at x1 = 42.2 mm and in b it is at x1 = 58.2 mm
400
Cs (GPa/m)
400
Cs (GPa/m)
Fig. 21 Plots showing the effects of impact velocity Vimp and interface stiffness parameter Cs on the mode of frictional sliding with a elastic bulk material behavior and b plastic bulk material behavior. The shaded symbols denote combinations of parameters that produce different frictional sliding modes in the elastic and plastic calculations. The initial compressive stress is 0 = 10 MPa and the reference flow strength for the plastic calculations is σ0 = 10 MPa
200
200
100
100
50
50 2 5
10
20
30
50
2 5
10
20
30
50
V imp (m/s)
V imp (m/s)
(a) elastic calculations
(b) plastic calculations
Legend: pulse-like mode
a train of pulses or majority of Δuslip is a train of pulses over the range of calculation
crack-like mode but with notable oscillations in slip-rate profile crack-like mode with relatively smooth slip-rate profile
123
66
Z. Shi et al.
calculations with Cs = 50 GPa/m, increasing Vimp tends to favor a crack-like mode or a train-of-pulses mode. Bulk material plasticity tends to smooth out oscillations which in turn tends to convert a train-ofpulses mode to a crack-like mode. The relatively small effect of bulk material plasticity on the frictional slip mode and on the propagation speed indicates that the interpretation of the Homalite on Homalite sliding experiments in Coker et al. (2005) based on elastic material behavior likely remains valid even if some bulk material plasticity did occur. One of the main motivations for exploring the mechanics of fast frictional sliding stems from earthquake dynamics. The loading conditions and bulk material constitutive relation used in the calculations here are quite different from those appropriate for modeling earthquake ruptures. Whether or not the effects of plasticity seen in the circumstances analyzed here carry over to the context of earthquake dynamics remains to be determined. Acknowledgments We are grateful to an anonymous referee for helpful comments and suggestions that have improved our paper. This work was supported by the USGS National Earthquake Hazard Reduction Program (grant G09AP00087) and the Southern California Earthquake Center (based on NSF Cooperative Agreement EAR-0106924 and USGS Cooperative Agreement 02HQAG0008).
(26)
with Cstan =
Cs 1 + γ t
(27)
∂u˙ slip ∂|Ts | C s
and
tan ˙ R p = Cs sgn(Ts ) u˙ slip (tn ) ∂u˙ slip ∂u˙ slip ˙ ˙ θ0 + θPC + γ t ∂θ0 ∂θPC
(28)
The partial derivatives of u˙ slip in Eqs. (27) and (28) are obtained by direct differentiation of Eq. (14) ∂u˙ slip r V0 β r − 1 = ∂|Ts | g(θ0 )θPC
(29)
∂u˙ slip −r V0 β r ∂g(θ0 ) = ∂θ0 g(θ0 ) ∂θ0
(30)
∂u˙ slip −r V0 β r = ∂θPC θPC where ∂g(θ0 ) = ∂θ0
(μs − μd ) θp0
(31)
L0 V1 θ0
p
p exp − VL1 θ00
Q 1/r L0 1 r V0 θ02 Q 1/r +1
(32)
with
The same rate tangent method as in Povirk and Needleman (1993), Coker et al. (2005) and Shi et al. (2008) is used to update T˙s . For completeness, the expressions used in the calculations are given here. The frictional sliding rate, u˙ slip , in [t, t + t] is expressed as a linear combination of its values at t and t + t u˙ slip = (1 − γ )u˙ slip (t) + γ u˙ slip (t + t), (24) where 0 ≤ γ ≤ 1. Expressing u˙ slip (t + t) as
123
T˙s = −[Cstan u˙ s − R˙ p ],
+ μ∗
Appendix: Rate tangent method for updating the interface shear traction
u˙ slip (t + dt) = u˙ slip (t)
∂u˙ slip ∂u˙ slip sgn(Ts )T˙s + + t θ˙0 ∂|Ts | ∂θ0
∂u˙ slip θ˙PC + ∂θPC
and substituting into Eqs. (12) and (24) gives an expression of the form
(25)
Q=
L0 +1 V0 θ0
L0 p μ∗ = μd + (μs − μd ) exp − V1 θ0
(33) (34)
References Ampuero J-P, Ben-Zion Y (2008) Cracks, pulses and macroscopic asymmetry of dynamic rupture on a bimaterial interface with velocity-weakening friction. Geophys J Int 173:674–692 Andrews DJ (1976) Rupture velocity of plane strain shear cracks. J Geophys Res 81:5679–5687 Andrews DJ (2005) Rupture dynamics with energy loss outside the slip zone. J Geophys Res 110:B01307. doi:10.1029/ 2004JB003191 Ando R, Yamashita T (2007) Effects of mesoscopic-scale fault structure on dynamic earthquake ruptures: Dynamic formation of geometrical complexity of earthquake faults. J Geophys Res 112:B09303. doi:10.1029/2006JB004612
Slip modes and partitioning of energy Beeler NM, Tullis TE (1996) Self-healing slip pulses in dynamic rupture models due to velocity dependent strength. Bull Seismol Soc Am 86:1130–1148 Belytschko T, Chiapetta RL, Bartel HD (1976) Efficient large scale non-linear transient analysis by finite elements. Int J Numer Meth Engr 10:579–596 Ben-Zion Y (2001) Dynamic ruptures in recent models of earthquake faults. J Mech Phys Solids 49:2209–2244 Ben-Zion Y, Shi Z (2005) Dynamic rupture on a material interface with spontaneous generation of plastic strain in the bulk. Earth Planet Sci Lett 236:486–496. doi:10.1016/j. epsl.2005.03.025 Brietzke GB, Cochard A, Igel H (2009) Importance of bimaterial interfaces for earthquake dynamics and strong ground motion. Geophys J Int 178: 921–938. doi:10.1111/ j.1365-246X.2009.04209.x Burridge R (1973) Admissible speeds for plane-strain self-similar shear cracks with friction but lacking cohesion. Geophys J Royal Astro Soc 35:439–455 Coker D, Lykotrafitis G, Needleman A, Rosakis AJ (2005) Frictional sliding modes along an interface between identical elastic plates subject to shear impact loading. J Mech Phys Solids 53:884–992 Day SM (1982) 3-dimensional finite-difference simulation of fault dynamics—rectangular faults with fixed rupture velocity. Bull seism Soc Am 72:705–727 Dalguer LA, Irikura K, Riera JD (2003) Simulation of tensile crack generation by three-dimensional dynamic shear rupture propagation during an earthquake. J Geophys Res 108(B3):2144. doi:10.1029/2001JB001738 Dieterich JH (1979) Modeling of rock friction 1. Experimental results and constitutive equations. J Geophys Res 84:2161– 2168 Duan B, Day SM (2008) Inelastic strain distribution and seismic radiation from rupture of a fault kink. J Geophys Res 113:B12311. doi:10.1029/2008JB005847 Festa G, Vilotte J-P (2006) Influence of the rupture initiation on the intersonic transition: Crack-like versus pulselike modes. Geophys Res Lett 33:L15320. doi:10.1029/ 2006GL026378 Freund LB (1998) Dynamic fracture mechanics. Cambridge University Press, Cambridge Heaton TH (1990) Evidence for and implications of self-healing pulses of slip in earthquake rupture. Phys Earth Planet In 64:1–20 Husseini MI (1977) Energy balance for motion along a fault. Geophys J R Astr Soc 49:699–714 Krieg RO, Key SW (1973) Transient shell response by numerical time integration. Int J Numer Meths Engrg 7:273–286 Linker MF, Dieterich JH (1992) Effects of variable normal stress on rock friction: observations and constitutive equations. J Geophys Res 97:4923–4940 Nielsen SB, Carlson JM, Olsen KB (2000) Influence of friction and fault geometry on earthquake rupture. J Geophys Res 105:6069–6088 Peirce D, Shih CF, Needleman A (1984) A tangent modulus method for rate dependent solids. Comp Struct 18: 875–887
67 Persson BNJ (2000) Sliding friction: physical principles and applications. 2nd edn. Springer, Berlin Poliakov ANB, Dmowska R, Rice JR (2002) Dynamic shear rupture interactions with fault bends and off-axis secondary faulting. J Geophys Res 107: B11, article no. 2295. doi:10. 1029/2001JB000572 Povirk GL, Needleman A (1993) Finite element simulations of fiber pull-out. J Eng Math Tech 115:286–291 Prakash V (1998) Frictional response of sliding interfaces subjected to time varying normal pressures. J Tribol 120:97– 102 Prakash V, Clifton RJ (1993) Pressure-shear plate impact measurement of dynamic friction for high speed machining applications. In: Proceedings of 7th International Congress on Experimental Mechanics, Society of Experimental Mechanics, Bethel, CT, pp 556–564 Rice JR, Ruina AL (1983) Stability of frictional sliding. J Appl Mech 50:343–349 Rice JR, Sammis CG, Parsons R (2005) Off-fault secondary failure induced by a dynamic slip pulse. Bull Seismol Soc Am 95:109–134 Rosakis AJ (2002) Intersonic shear cracks and fault rupture. Adv Phys 51:1189–1257 Rosakis AJ, Samudrala O, Coker D (2000) Intersonic shear crack growth along weak planes. Mater Res Innov 3:326–343 Ruina AL (1983) Slip instability and state variable friction laws. J Geophys Res 88:10359–10370 Samudrala O, Huang Y, Rosakis AJ (2002) Subsonic and intersonic shear rupture of weak planes with a velocity weakening cohesive zone. J Geophys Res 107:B8, article no. 2170. doi:10.1029/2001JB000460 Shi Z, Ben-Zion Y (2006) Dynamic rupture on a bimaterial interface governed by slip-weakening friction. Geophys J Int 165:469–484. doi:10.1111/j.1365-246X.2006.02853.x Shi Z, Ben-Zion Y, Needleman A (2008) Properties of dynamic rupture and energy partition in a solid with a frictional interface. J Mech Phys Solids 56:5–24 Templeton EL, Rice JR (2008) Off-fault plasticity and earthquake rupture dynamics: 1. Dry materials or neglect of fluid pressure changes. J Geophys Res 113:B09036. doi:1029/ 2007JB005529 Viesca RC, Templeton EL, Rice JR (2008) Off-fault plasticity and earthquake rupture dynamics: 2. Case of saturated offfault materials. J Geophys Res 113:B09307. doi:10.1029/ 2007JB005530 Weertman JJ (1980) Unstable slippage across a fault that separates elastic media of different elastic constants. J Geophys Res 85:1455–1461 Xia K, Rosakis AJ, Kanamori H (2004) Laboratory earthquake: the sub-Rayleigh-to-supershear rupture transition. Science 303:1859–1861 Yamashita T (2000) Generation of microcracks by dynamic shear rupture and its effects on rupture growth and elastic wave radiation. Geophys J Int 143:395–406 Zheng G, Rice JR (1998) Conditions under which velocityweakening friction allows a self-healing versus a crack-like mode of rupture. Bull Seismol Soc Am 88:1466–1483
123
