Access the full text.
Sign up today, get DeepDyve free for 14 days.
Loading next page...
/lp/arxiv-cornell-university/modeling-wavefields-in-saturated-elastic-porous-media-based-on-CD5dmE0JR0
- ISSN
- 0045-7930
- eISSN
- ARCH-3341
- DOI
- 10.1016/j.compfluid.2020.104587
- Publisher site
- See Article on Publisher Site
Abstract
Modeling wave elds in saturated elastic porous media based on thermodynamically compatible system theory for two-phase solid- uid mixtures 1 1; 2 3; 4 3 Evgeniy Romenski, Galina Reshetova, Ilya Peshkov, Michael Dumbser June 11, 2020 Abstract A two-phase model and its application to wave elds results of solving some numerical test problems are presented numerical simulation are discussed in the context of modeling and discussed of compressible uid ows in elastic porous media. The deriva- tion of the model is based on a theory of thermodynamically 1 Introduction compatible systems and on a model of nonlinear elastoplastic- ity combined with a two-phase compressible uid ow model. The modeling of uid ows in porous media is of permanent The governing equations of the model include phase mass con- interest in many geophysical and industrial applications. The servation laws, a total momentum conservation law, an equa- starting point of research developments in this eld was a se- tion for the relative velocities of the phases, an equation for ries of pioneering works by Biot [2, 1, 3], in which a model of mixture distortion, and a balance equation for porosity. They elastic wave propagation in saturated porous media was pro- form a hyperbolic system of conservation equations that sat- posed. Some modi cations and generalizations of the model isfy the fundamental laws of thermodynamics. Two types of have been made (see, for example [6, 21, 46] and references phase interaction are introduced in the model: phase pressure therein), and at present Biot's approach is a commonly ac- relaxation to a common value and interfacial friction. Inelas- cepted and widely used one in geophysical community. Nev- tic deformations also can be accounted for by source terms ertheless, many actual technological and scienti c problems, in the equation for distortion. The thus formulated model such as geothermal energy extraction, CO2 storage, hydraulic can be used for studying general compressible uid ows in fracturing, fuel cells, food production, etc. require new ad- a deformable elastoplastic porous medium, and for modeling vanced models and methods. wave propagation in a saturated porous medium. Governing To simulate the development of nonlinear, temperature de- equations for small-amplitude wave propagation in a uniform pendent processes in porous media, methods of continuum porous medium saturated with a single uid are derived. They mechanics and, in particular, multiphase theories can be suc- form a rst-order hyperbolic PDE system written in terms of cessfully used. A two-phase approach to poroelasticity has stress and velocities and, like in Biots model, predict three type been, perhaps, most consistently implemented by Wilmanski of waves existing in real uid-saturated porous media: fast and in [45, 44] (see also references therein). In particular, in a re- slow longitudinal waves and shear waves. For the numeri- view paper, [44], the structure of Biot's poroelastic model was cal solution of these equations, an ecient numerical method analysed and its consistency with the fundamental principles based on a staggered-grid nite di erence scheme is used. The of continuum mechanics was discussed. Sobolev Institute of Mathematics, 4 Acad. Koptyug Avenue, 630090 In recent years, considerable attention has been paid to the Novosibirsk, Russia, e-mail: evrom@math.nsc.ru modeling nite-strain saturated porous media and their appli- Institute of Computational Mathematics and Mathematical Geo- cations in various elds, in particular in medicine, see, for ex- physics, 6 Pr. Akademika Lavrentjeva, 630090 Novosibirsk, Russia, e- mail: kgv@nmsf.sscc.ru ample, [19, 34, 29]. Worthy of mention is a large-deformation Department of Civil, Environmental and Mechanical Engineering, model of a saturated porous medium proposed by Dorovsky University of Trento, Via Mesiano 77, 38123 Trento, Italy, e-mail: [4], in which the key point of the model is thermodynamic con- ilya.peshkov@unitn.it sistency and hyperbolicity of the governing equations. Nev- The work by I.P. has been started while being at Paul Sabatier University, Institut de Mathe ematiques de Toulouse, Toulouse, France ertheless, there is still no thermodynamically consistent for- arXiv:1910.04207v2 [physics.flu-dyn] 9 Jun 2020 mulation of a multiphase mixture ow model in a deforming continuum mechanics and for a two-phase compressible uid porous medium with nite deformations. model. By combining the above two models, a governing mas- ter SHTC system for a two-phase solid- uid medium is for- In this paper, we apply a powerful method of designing mulated. In Section 3, using this model governing equations new models of complex continuum media, which is based on are derived for small-amplitude wave propagation in a sta- a thermodynamically compatible system theory [12, 14, 13]. tionary saturated porous medium. In Section 4 we discuss the In [35], the mentioned theory was rst time applied to de- di erences and similarities between the Biot and SHTC mod- signing a model for deformed porous medium saturated by els. In Section 4.2, we derive a dispersion relation for the thus a compressible uid. In this paper the governing equations obtained acoustic equations and study the properties of the were formulated and small amplitude wave propagation has wave elds. In Section 5, an ecient nite di erence method been studied showing a qualitative agreement with Biot's the- for small-amplitude wave propagation is presented and some ory. The present study contains further developments of ideas numerical results are discussed. from [35] based on the uni ed model of continuum [9, 10] as well as a quantitative comparison with Biot's theory and a se- ries of numerical experiments proving the physical correctness 2 System of governing PDEs for of the model. The thermodynamically compatible system the- poroelastic media ory allows the development of well-posed models satisfying the fundamental laws of irreversible thermodynamics. In [14, 38, To develop a poroelastic model whose governing equations 40, 15, 32], a class of Symmetric Hyperbolic Thermodynami- form a symmetric hyperbolic thermodynamically compatible cally Compatible (SHTC) systems was formulated to describe (SHTC) system, a uni ed thermodynamically compatible con- many known classical equations of continuum mechanics and tinuum model formulated in [31, 9] is coupled to a two-phase electrodynamics ( uid mechanics, solid mechanics, electrody- compressible uid model [36]. namics, magnetohydrodynamics) including advective and dis- sipative processes. All SHTC systems have nice mathematical 2.1 SHTC governing equations of the uni ed properties: symmetric hyperbolicity in the sense of Friedrichs [11] and a conservative form of the equations. The solutions model of continuum to the governing PDE system satisfy fundamental laws of non- We start our consideration with a brief recall of the uni ed equilibrium irreversible thermodynamics: conservation of to- model of continuum, the formulation of which can be found tal energy ( rst law) and non-decreasing of physical entropy in [31, 9]. This model can describe elastic and elastoplastic (second law). behavior of the medium, as well as ows of a viscous uid by The SHTC system theory is a rst principle type theory. It a single PDE system. First, we recall the governing equations allows the derivation of governing PDEs for a quite wide class of the uni ed model for a deforming and owing continuum of physical processes from a variational principle [32] (Hamil- which reads ton's principle of stationary action). In particular, the SHTC i k approach has been successfully applied to the development of @ v v + p @v ik ik + = 0; (1a) a hierarchy of compressible multi-phase ow models [37, 36, @t @x 39, 30]. Recently, a uni ed SHTC model of Newtonian contin- @ @v uum mechanics has been developed [9, 10]. It simultaneously + = 0; (1b) @t @x describes the dynamics of elastoplastic solids, as well as of viscous and non-viscous uids in the presence of electromag- @A @A v @A @A ik ij ik ij ik netic elds. In the present paper, we extend this uni ed model + + v = ; (1c) @t @x @x @x k j k to describe solid- uid two-phase ows. The governing equa- tions of the model also belong to the class of SHTC equations. @s @sv The interfacial friction between the liquid and solid phases + = 0: (1d) ik ik @t @x T and shear stress relaxation are implemented in the model as relaxation-type source terms in accordance with the laws of Here, (1a) is the linear momentum conservation law, (1b) is thermodynamics. The latter allows taking into account the the mass conservation law, (1c) is the evolution for the distor- time dependence of elastic moduli on frequency and using nu- tion matrix, and (1d) is the entropy balance law. merical methods to study wave propagation problems. As an independent set of state parameters we take the The rest of the paper is organized as follows. Section 2 medium velocity v , mass density , distortion A , and en- ik brie y describes the governing PDEs for a uni ed model of tropy s. Note that positioning of the tensorial index in the 2 k velocity eld v as a superscript is merely conditioned by that reads as the subscript is reserved for the constituent index and is not 2 2 c tr(g ) sh 2 intending to distinguish covariant and contravariant compo- = g I ; tr() = 0: (6) 2 3 nents of a tensor which coincide with each other in a Carte- sian coordinate system. The density is connected with the The coecient is a function of parameters of state, and it distortion by an algebraic compatibility constraint such as can be taken in the form [9, 8, 33] = detA, where is a reference density. Additional pa- 0 0 rameters of the medium presented in the above system are c (; T; Y )>0; (7) sh pressure p, shear stress , and temperature T . They are ik connected with density, distortion, and entropy via speci c where Y = tr( ) is the intensity of shear stress. total energy E(v; ; s;A): The source term in the equation for distortion produces a @E @E 2 2 nonnegative entropy production source term in (1d). It is p := E ; := A A E ; ij ki ki A kj @ @A kj important to note that for system (1) there is an additional @E conservation law [32]: T := E : (2) @s k i @E @ v E + v (p ) ik ik + = 0; (8) The source term in the equation for distortion characterizes @t @x @E the rate of inelastic deformation, where = , while the @A which is the conventional energy conservation law. function ( ) depends on the shear stress relaxation time . This strain relaxation source term is one of the key elements of the model since it makes it possible to apply the model to 2.2 SHTC governing equations of two-phase description of both elasto-plastic deformations of solids [15, 33] as well as ows of viscous uids [9, 8, 18, 5]. compressible uid ow model In the two-phase compressible uid model, the ow is consid- To close the model, we de ne how the energy E and the ered as a mixture of two immiscible constituents with their parameter depend on the parameters of state. We take the own parameters of state. Thus, the general model should take energy in the form into account the di erence between velocities, pressures, and temperatures of the constituents. That means that if we con- E = E (; s) + E (; s;A) + E (v); (3) 1 2 3 sider velocity, density, and entropy to be the basic parameters of state, they can be di erent for each phase. In our consider- where E = kvk is speci c kinetic energy, E is the "hydro- 3 1 ation, we restrict ourselves to a single entropy approximation dynamic" part corresponding to the energy of volume defor- for small variations of phase temperatures. In [39], it is proved mations only, and E is the energy of shear strain. In order that for multiphase compressible mixtures a single entropy to provide a zero trace of shear stress, tr() = 0, we take E approximation is suitable for the ows close to thermal equi- depending on A via the normalized strain tensor g = a a, 1=3 1=3 librium. In this case, the change of phase temperatures due where a = A=(detA) . It gives g = G=(detG) , where to the small variation of phase entropies is negligibly small. G = A A. Then, energy E can be de ned as in [25]: Finally note, that we consider only small density variations 2 2 which means that the processes under consideration are close E = c tr(g ) 3 ; (4) sh to thermal equilibrium. The SHTC system for two-phase compressible ow with a where c is the shear velocity of sound under reference con- sh single entropy approximation [36] reads as follows: ditions. i i k i @v @(v v + p + w E k ) ik w + = 0; (9a) Using the above de nition, we can take the derivative of E @t @x with respect to A: @ @v 2 2 + = 0; (9b) @E @E c tr(g ) 1 T sh T 2 @t @x = = F = F g I ; (5) k @A @A 2 3 @c @(c v + E ) 1 1 w 1 + = 0; (9c) where F = A . Then the shear stress is trace-free, and it @t @x k l l k l @w @(w v + E ) @w @w c k modynamic forces (the derivatives of internal energy with re- 1 l + + v = ; (9d) @t @x @x @x spect to the parameters of state) [36, 39]: k l k 2 p p @ @ v 2 1 1 1 2 E = ; p = E = p + p ; E = c c w ; + = ; (9e) 1 1 2 2 w 1 2 i 1 i @t @x k 1 E = e + p = e p = + (1 2c )kwk =2; c 1 1 1 2 2 2 1 @s @sv + = + 0; (9f ) k k @e @e 1 2 @t @x T T k 1 2 E = T = c + c : (13) s 1 2 @s @s Here, is the volume fraction of the rst phase, which is @e 2 i Here p = ; (i = 1; 2) is the phase pressure. connected with the volume fraction of the second phase via @ In [39], a model of multiphase compressible ow with ar- the saturation law + = 1, is the mixture mass density, 1 2 bitrary number of phases is presented, and it is shown that which is connected with the phase mass densities ; via 1 2 its governing equations can be written as a symmetric hyper- the relation = + . The phase mass fractions are 1 1 2 2 bolic system. It means that the two-phase equations (9) with de ned as c = =; c = = (c +c = 1). Eventually, 1 1 1 2 2 2 1 2 i i i i i i the closing relations (13) can also be written as a symmetric v = c v + c v is the mixture velocity, w = v v is 1 2 1 2 1 2 hyperbolic system. the phase relative velocity, s is the speci c entropy of the mixture, and as throughout this paper, we use the notations @E @E E k , E , etc. k c w 1 @w @c 2.3 Thermodynamically compatible master The phase interaction is presented as algebraic source terms system for two-phase saturated porous in (9d) and (9e) which are proportional to thermodynamic media forces. These source terms are phase pressure relaxation to a common value and interfacial friction: = = E = 1 1 The SHTC systems of the uni ed continuum model and the and = = E k= . The coecients ; characterize k 2 w 2 1 2 two-phase compressible uid ow model presented in the pre- the rate of pressure and the velocity relaxation. They can vious sections can be obtained as subsystems of a general mas- depend on the parameters of state. ter system written with the use of a generalized internal en- The entropy production in the entropy balance equation ergy. This master system can be derived from the rst prin- (9f) is non-negative due to the de nition of the phase interac- ciples by minimizing the Lagrangian and passing to Eulerian tion source terms. The energy conservation law holds in the coordinates, see [32]. Its derivation and connection with a following form: Hamiltonian formulation of irreversible non-equilibrium ther- modynamics known as GENERIC is discussed in [32]. In [30], k i i @E @ v E + v (p + w E k ) + E E ik c w w 1 k the above-mentioned master system is applied to the deriva- + = 0; (10) @t @x tion of a model for two-phase solid- uid media experiencing a stress-induced solid- uid phase transformation. The goal of where E = E ( ; c ; ; s) + E (v) + E (c ;w), E is the spe- 1 1 1 3 4 1 1 the present paper is to develop a two-phase model for com- ci c internal energy of the mixture, E = kvk =2 is the kinetic pressible uid ow in deforming porous media. 1 2 energy of the mixture, and E = c c kwk is the kinematic 4 1 2 Let us consider the following PDE master system: energy of the relative motion. Note that i i k i @v @(v v + p + w E k ) ik ik 2 2 + = 0; (14a) E + E = ( kv k + kv k ): (11) 3 4 1 1 1 2 2 2 @t @x @A @A v @A @A ik im ik ij ik The most important closing relation for system (9) is the + + v = ; (14b) @t @x @x @x k j k internal energy E , which should be chosen in such a way that the governing equations take the form of the well-known @ @v balance laws of the two-phase ow model. In [36, 39] it is + = 0; (14c) @t @x de ned as the mass averaged phase equation of state @c (@c v + E k ) 1 1 w + = 0; (14d) E = c e ( ; s) + c e ( ; s); (12) 1 1 1 1 2 2 2 @t @x where e ( ; s) is the speci c internal energy of the i-th phase. k l l k l i i @w @(w v + E ) @w @w c k 1 l + + v = ; (14e) This de nition leads to the following formulae for the ther- @t @x @x @x k l k 2 4 k @ @ v ' 2 1 1 E = c c kwk : (17) + = ; (14f ) 4 1 2 @t @x k 1 k We take the part of energy related to volume deformation as @s @sv + = + ' + 0; (14g) ik ik k k the mass averaged energy of the phases, because it naturally @t @x T T T k 1 2 follows from the additivity of the energy per unit volume: The application of the master system (14) to designing E = e ( ; s) + e ( ; s): (18) 1 1 1 1 1 2 2 2 2 a concrete physical process has to be done by the proper choice of the generalized energy potential E. If, for example, The kinetic energy of the relative motion, E , remains the we take E = E(v; ;A; s), we can neglect equations (14d), same as is in the two-phase ow model. (14e), (14f) and obtain the PDEs for the uni ed model of There is no rigorous justi cation for determining the shear continuum mechanics (1). On the other hand, if we take strain energy, E . It is clear that the volume deformation E = E( ; c ; ;w; s), we obtain the governing PDEs for com- 1 1 of two-phase mixtures can be represented as the sum of the pressible two-phase ow (9). Therefore, the governing PDEs volume deformations of the solid and uid constituents, that for deformed continuum and for compressible two-phase uid is why we de ne the volumetric energy as the mass averaged ow can be viewed as consequences of (14). It seems natural to phase energies. As for any arbitrary strain measure of the take this general system as a basis for a two-phase solid- uid mixture element, it is not clear whether it can be divided into mixture model. parts related to the solid and uid constituents separately. First, we identify the parameters of state in (14) with phys- That is why we use the distortion of the mixture as a pa- ical parameters characterizing deforming porous media. Let rameter of state, the rate of which is connected with spatial the parameter characterize the volume fraction of the uid derivatives of the mixture velocity. Moreover, we assume that component in the solid- uid mixture, which means that it can the shear part of internal energy depends on the elastic defor- be identi ed with the porosity, usually denoted by . Then mation tensor of the mixture and de ne it by formula (2.1): 1 2 2 = 1 is the volume fraction of the solid phase of the 2 1 E = c tr(g ) 3 ; where c is the shear speed of 2 sh;m sh;m porous material. As in Section 2.2, the mixture density is sound of the mixture to be determined. It can be naturally connected with the phase mass densities via = + . 1 1 2 2 assumed that in the limiting case c = 0; c = 1, which corre- 1 2 The parameter c represents the mass fraction of the uid sponds to the pure solid state, the shear energy corresponds component. And, by analogy with the two-phase ow model, to the elastic shear energy if c equals the shear sound ve- sh;m locity of the solid medium. On the other hand, recalling that i i i i i i v = c v + c v ; w = v v (15) 1 2 1 2 1 2 we consider the ideal uid (no resistance to shear), we can take c = 0 in the pure uid limiting case (c = 1; c = 0) sh;m 1 2 is the velocity of the mixture and the relative velocity, re- and hence, there is no contribution to the shear part of the spectively. We consider a single entropy approximation of the total energy. From this reasoning, it follows that for the in- two-phase medium and introduce the entropy of the mixture s. termediate case (0 < c ; c < 1; ), we should choose c in 1 2 sh;m As noted at the beginning of Section 2.2, for processes which such a way that it disappears for the pure uid medium and are close to thermal equilibrium, the single entropy approxi- is equal to the solid shear speed of sound for the pure solid mation is acceptable. Finally, as a measure of deformation of medium. That is why we assume that for the solid- uid mix- the element of the porous medium we consider the distortion 2 2 ture c = c c , where c is the shear speed of sound of 2 sh sh;m sh A of the mixture. the skeleton, and take the shear energy in the form Let us now take the generalized total energy potential in the same form as in the case of a uni ed deforming continuum (3), 2 2 E = c c tr(g ) 3 : (19) 2 2 sh but taking into account the two-phase nature of the medium 8 and assuming that the energy corresponding to volume defor- It can be shown then via the characteristic analysis [35] that mation is de ned as in the two-phase uid model: such a choice of the shear elastic energy gives the shear char- acteristic speed for the mixture c = c c , that seems to sh;m 2 sh E = E ( ; c ; ; s) + E (c ; ; s;A) + E (v) + E (c ;w): 1 1 1 2 1 3 4 1 be a reasonable approximation but can be further re ned if (16) it is necessary to achieve better agreement with experimental data. Here we assume that As soon as we de ne the generalized energy by (16), (17), E = c e ( ; s) + c e ( ; s); E = kvk ; 1 1 1 1 2 2 2 3 (18), (19), the closing relations in the phase interaction source 5 2 p p kwk terms are de ned via the thermodynamic forces: 1 2 E = e + e c E + (1 2c ) ; c 1 2 2 1 1 2 1 2 = E ; = E ; ' = E : (20) ik A k w 1 ik k 1 2 2 E = c c tr(g ) 3 ; 2 2 sh @e @e 1 2 The solution to system (14) satis es the energy conservation E = T = c + c : (23) s 1 2 @s @s law which reads as For the derivation of (23) we use formula (18) for the in- @E @ k i i + v E + v (p + w E k ) ternal energy of the mixture and the relationship between ik w ik @t @x the mixture parameters and the individual phase densities: c c 1 2 +E E ) = 0 (21) c w 1 k = ; = . Details of the derivation can be found, for 1 2 1 2 example, in [36, 39]. and should be used instead of the entropy balance law (14g) System (22) is equivalent to (14) if the total mass conser- in the development of numerical methods. vation law (14c) is replaced by the mass conservation law for the second phase (22d). With the above de nitions of the energies, we obtain the following system of governing PDEs for compressible ow in deforming porous media, which is written in terms of the phase 3 System of governing PDEs for parameters: small-amplitude wave propagation in i i @( v + v ) @ 1 1 2 2 1 2 i k i k + v v + v v 1 1 2 2 1 1 2 2 saturated porous media @t @x + p = 0; (22a) ik ik In this section, we derive equations for small-amplitude wave j propagation in a saturated porous medium at equilibrium. For @A @A v @A @A ik ij ik ij ik + + v = ; (22b) the derivation, we rst transform equations (22) to a more @t @x @x @x k j k convenient form. Instead of the total momentum equation and the equation for the relative velocity, consider momentum @ @ v 1 1 1 1 + = 0; (22c) equations for each of the constituents. These are derived from @t @x (22a) and (22e) and read as @ @ v 2 2 2 2 + = 0; (22d) i i @v @v 1 @p 1 @ s p p @ 1 2 ik 1 2 1 1 k 1 @t @x + v + + @t @x @x @x @x k 1 i k i j j j j p p 1 1 2 i k k (v v v v ) + e + e E 1 2 2 @A @v @v 1 1 2 2 @w 2 mn 1 2 k k 1 1 c a + c (v v ) c + 2 mn 2 2 1 2 @x @x @x @t @x i k i k l i k @v @v c c @w @w k 2 2 1 2 2 2 i i +c = (v v ); (24) +v = ; (22e) 1 1 2 @x @x @x @x k i 2 l k 2 @ @ v ' 1 1 + = ; (22f ) i i @v @v 1 @p 1 @ s p p @ 2 2 ik 1 2 1 2 k 2 @t @x k 1 + v + + @t @x @x @x @x k 2 i k i @s @sv i k @A @v @v + = + ' + 0: (22g) mn k k 1 1 ik ik k k +c a c (v v ) c @t @x T T T 1 mn 1 2 k 1 2 1 2 @x @x @x i k i i k For the derivation of the above system from equations (14), @v @v c c 1 1 1 2 2 2 i i +c = (v v ); (25) 1 2 the following formulae for the thermodynamic forces are used: @x @x k i 2 p p 2 1 1 where a = @(c E )=@A and the derivative @E =@A E = ; p = E = p + p ; mn 2 mn 2 mn 1 1 2 2 2 can be taken as in (5). c 1 sh In addition to the above equations, we consider phase mass = s ; s = (g g g g ); ij 2 ij ij ik kj mn nm ij 2 3 conservation laws and balance equations for the volume frac- E i = c c w ; tion, distortion, and entropy, which can be written in the fol- w 1 2 6 lowing quasilinear form: used: k K K 1 2 @ @ @v c @ 1 1 1 2 1 k 1 k k p = ; p = ; (30) 1 1 2 2 + v + + (v v ) 0 0 1 1 2 1 2 @t @x @x @x k k 1 k p p 1 2 1 @p @p = ; (26a) 0 1 0 2 where K = ; K = are the 1 2 1 2 @ @ 1 1 1 0 2 0 = ;s=0 = ;s=0 1 2 1 2 bulk moduli of the uid and solid phases, respectively. @ @ @v c @ 2 2 2 2 1 1 k k k + v + + (v v ) The deformation of the medium in case of small perturba- 2 1 2 @t @x @x @x k k 2 k tions of the equilibrium solution can be described by any of p p 2 2 1 the known equivalent strain tensors. In particular, one can use = ; (26b) 2 1 T the Almansi strain tensor " = (I G)=2, where G = A A. For small perturbations of distortion A = + A , the @ @ p p ij ij ij 1 1 1 2 + v = ; (26c) stress-strain relation (23) takes the form @t @x k 1 " + " + " @A @A @v E 11 22 22 ij ij A 2 k ik = 2 c " ; (31) + v + A = ; (26d) ij 2 2 ij ij im sh @t @x @x k j @s @s 1 1 (p p ) where " = (A + A )=2 are the components of the Al- ij ij ji 1 2 + v = E E + A A ik ik mansi strain tensor ". Thus, for small perturbations we have @t @x T T k 1 " + " + " + E kE k: (26e) 11 22 22 w w 0 0 0 2 = s = 2 c " : (32) 2 ij ij ij ij 2 2 2 sh Small amplitude wave propagation can be described by a Small perturbation of E = F reads as PDE system obtained by linearization of (26) with coecients 0 2 de ned in the equilibrium state of the original system. Assume c " + " + " 11 22 22 0 sh E = = 2 " : (33) A ij ij that a medium with given volume fractions of the constituents ij 2 0 0 is at rest and under reference conditions. This means that its parameters of state are Now, omitting the notation for the perturbations, we ob- tain the following system: k k 0 0 0 v = v = 0; = ; = ; = ; 1 2 1 1 2 1 2 1 i 0 0 0 0 0 0 0 @v 1 @p @s c c 1 ik = = 1 ; A = ; s = 0: (27) 1 2 2 2 1 2 i i 2 ij ij 2 1 + = (v v ); (34a) 1 2 0 0 0 @t @x @x i k 2 The other parameters of the medium computed with the above i 0 0 0 0 0 @v 1 @p @s c c 2 ik 2 2 1 1 1 2 i i ones also correspond to the medium at rest: + = + (v v ); (34b) 1 2 0 0 @t @x @x i k 2 k k k 0 p = p = 0; s = 0; v = 0; w = 0; a = 0: (28) @ @v p p 1 2 1 1 2 ik mn 1 1 + = ; (34c) @t @x k 1 We are interested in di erential equations for small pertur- k 0 @ @v p p 2 2 1 0 2 2 + = ; (34d) bations of the equilibrium solution (27). Thus, our goal is to 0 0 @t @x k 1 nd a solution to the system of equations (24), (25), and (26) @ p p 1 1 2 in the form = ; (34e) @t k k k k 0 m @A @v E v = 0 + v ; v = 0 + v ; = + ; 1 1 ij A 1 1 2 2 1 ik + = (34f ) im @t @x = + ; (29a) 2 2 @s = 0: (34g) = + ; A = + A ; s = 0 + s: (29b) 1 1 ij ij ij 1 @t Equations for perturbations of the equilibrium solution can One can see that entropy remains constant in time. That be derived by substituting (29) into (24), (25), and (26) and means that entropy has no in uence on small perturbation neglecting the terms of orders higher than the rst one. The wave elds, and the equation for entropy can be neglected. following relations for perturbed phase pressures will also be Also, because the shear stress s depends on the small strain ij 7 tensor " , the equation for distortion A (34f) can be replaced can be replaced by a single equation for the pressure P : ij by an equation for ", which reads as 0 k 0 k @P @v @v 1 1 2 2 + + = 0: 0 1 0 1 0 1 0 1 i j @t @x @x @" 1 @v @v K + K K + K k k ij 1 1 2 2 1 1 2 2 (37) @t 2 @x @x j i 0 2 2 c " + " + " 11 22 22 0 2 sh k k 0 k = " : (35) Now, taking into account (31), (37) and v = v + c w , ij ij 2 1 2 k k 0 k v = v c w , we obtain the resulting system written in 2 1 terms of the mixture velocities, relative velocities, pressure, and shear stress: Finally, we obtain the following system: i 0 0 0 0 0 @v @P @s ik 0 0 @v 1 @p @s c c 1 ik 1 2 2 2 1 2 i i + = 0; (38a) + = (v v ); (36a) 1 2 0 @t @x @x 0 0 i k @t @x @x i k 2 k 0 0 @w 1 1 @P c c i 0 0 0 0 0 1 2 k @v 1 @p @s c c 2 ik 2 2 1 1 1 2 i i + = w ; (38b) 0 0 + = + (v v ); (36b) 1 2 @t @x 0 0 0 i 2 1 2 @t @x @x i k 2 k 0 0 k k 0 @P @v @w 1 2 0 0 @ @v p p 1 2 1 0 1 1 + K + K = 0; (38c) 2 1 + = ; (36c) 0 0 @t @x @x 0 k k @t @x k 1 i k j @s @v @v 2 @v s k 0 ik ik @ @v p p 2 2 1 0 2 2 + = (38d) ik + = ; (36d) 2 @t @x @x 3 @x 0 0 k i j @t @x k 1 @ p p 1 1 1 2 1 1 0 0 0 2 where K = K + K , = c , and is the = ; (36e) 1 1 2 2 2 sh @t shear stress relaxation time. i j 0 2 @" 1 @v @v 2 c ij 0 2 sh Thus, we have derived the PDE system (38) for small am- + = " ij @t 2 @x @x j i plitude wave propagation in a saturated porous medium at rest. In this system, two dissipation mechanisms are present: " + " + " 11 22 22 : (36f ) ij (i) friction of the uid on the pore walls and (ii) shear stress relaxation of the saturated porous medium. In the following section, some properties of the wave elds governed by (38) These equations form a basis for our study of small- will be studied. amplitude waves in a porous saturated medium at rest. Furthermore, we will impose a condition that will help fur- 4 Comparison of SHTC and Biot models ther simplify the PDE system (36). Our concern is to derive equations for waves which have wavelengths much bigger than 4.1 Theoretical comparison the characteristic size of the pores. This means that we assume instantaneous phase pressure equalizing. This is because the The most comprehensive comparative study of Biot's model process is fully determined by pressure waves propagation and and models developed by a classical two-phase approach based re ection at the pore boundaries and, thus, the characteristic on continuum thermodynamics has been made by Wilmanski time scale for reaching a pressure equilibrium is small. [44]. The conclusion of this paper is that the two-phase model contains all information in Biot's model about the features of The instantaneous relaxation of the phase pressures leads to wave propagation in a saturated porous medium, and, in par- governing equations obtained in the relaxation limit of (36) as ticular, it predicts the slow pressure waves and gives a qual- ! 0. In this case the resulting system is the system (36) in itatively correct description of the dependence of the phase which the equation for is replaced by an algebraic equation: velocities on frequency. p = p = P . Then, since the phase pressures are equal, we 1 2 The governing equations of the SHTC model proposed in have from (30) that the present paper di er from those obtained by the classical two-phase solid- uid approach, but we will see that qualita- K K 1 2 1 2 0 0 tively the properties of wave elds are the same as those in 1 2 Biot's model. We emphasise that the above-proposed SHTC and, hence, the two equations for the phase densities ; model can be used for describing the nite deformations in 1 2 8 i i 0 0 the medium by taking into account changes in porosity under v = v , v = v for the velocities, = , = for the f s f s 1 2 1 2 stress variations and inelastic deformations. It should be also densities, and c = c , c = c for the mass fractions, equation 1 f 2 s emphasized that the consideration of nite deformations im- (38b) for the relative velocity can be replaced by the following poses certain restrictions on the de nition of deformations of equivalent equation for the momentum of the uid: the constituents. For example, the density of the mixture is an i 2 @v @P @ c f ik f additive parameter which can be de ned as the sum of partial f s i i + = (v v ); (42) f s @t @x @x densities, = + . On the other hand, the decom- i 0 k 2 1 1 2 2 position of arbitrary nite deformation into deformations of where = + is the mixture density, P is the uid f f s s the solid and uid phases is questionable, because this could pressure, = s is the shear stress, and c ; c are the mass ik s ik f s imply the existence of two relaxed reference frames and two fractions of the uid and solid constituents, respectively. The sets of Lagrangian coordinates. That is why we consider dis- shear stress gradient in the left hand-side of (42) is a term tortion A as a measure of deformation of the whole medium which is present in the SHTC equations and absent in Biot's and do not consider deformation of the skeleton separately. In model, e.g. see equation (43a) below. In fact, from a physical this context, we recall that our uni ed continuum model for standpoint, taking into account the shear stress in the balance uids and solids [31, 9, 8] also relies on a deformation-based equation for the uid momentum seems to be rather natural, rather than a strain-rate-based description of uid ows. because the motion of the uid may be caused not only by the The stress-strain relation in the SHTC model fully depends pressure gradient, but also by tangential deformation of the on the de nition of the thermodynamic potential E (16). The mixture element. total stress which is presented in the total momentum equation Overall, the equations for the pressure and shear stress in (22a) reads as the SHTC and Biot's models are quite di erent and cannot be transformed one into another, because the stress-strain re- T = ( p ) p ; (39) ik 2 ik 2 1 1 lationships and elastic moduli used in these models are dif- ferent. Nevertheless, it will be shown below that the features where is the deviatoric trace-free stress tensor (tr() = 0). ik of wave elds are qualitatively similar in both models, and in In Biot's model, the total stress is de ned as some cases they are quantitatively close by a corresponding ef choice of the material parameters. In the following section, T = nP ; (40) ik ik ik this statement will be discussed in a more quantitative man- ef ner via a plane wave analysis. where is the so-called e ective stress, P is the pore pres- ik sure, and n is the Biot coecient (see, for example, [22]). ef Thus, if we take = in our SHTC formulation (39), 2 ik ik 4.2 Analysis of wave elds in saturated porous we have media ef T = p p ; (41) ik 1 1 2 2 The 3D system of governing equations for small-amplitude ik wave propagation (38) is quite complex, but an informative and in the case of instantaneous pressure relaxation with p = analysis can be done for 1D plane waves. In this section, p = P , we obtain Biot's formula (40) with n = 1. we neglect the stress relaxation term in the equation for the Furthermore, note that there is a possibility to specify the shear stress tensor of the SHTC model, as it is absent in Biot's energy potential in such a way that formula (40) holds also in model. the SHTC model. To do this we de ne the volumetric part of internal energy as 4.2.1 1D Biot's theory E = nc e ( ; s) + e ( ; s) The one-dimensional rst-order form of the governing equa- 1 1 1 1 1 1 2 2 2 2 tions of the so-called \low-frequency limit" in Biot's theory instead of (18), and take the pressure relaxation term in the for the vector of state variables (v ; q; ; p) can be formulated np p 1 2 equation for volume fraction as . In this paper we do as follows [6, 21]: not discuss this possibility in detail. @v T @q @p As to the momentum balance for the uid constituent in + + = q; (43a) f f @t @t @x the SHTC model, there is an additional force to induce uid ow, which is connected with the shear stress gradient and @v @q @ + = 0; (43b) not presented in Biot's model. Speci cally, with the notation f @t @t @x 9 0 1 1 1 @ @v @q s 0 0 0 0 0 0 0 0 ( + 2 ) M = 0; (43c) u u B C @t @x @x 0 0 0 0 0 0 0 0 0 B C B C 0 0 0 0 0 0 0 0 0 B C @p @v @q B C + M + M = 0; (43d) 0 0 0 0 0 0 R 0 0 0 B C @t @x @x B C 0 0 0 0 0 0 0 0 0 0 B C B C 0 0 0 0 0 0 0 0 0 0 where is the porosity, v and q = (v v ) are the solid and s f s B C B C K 0 0 K 0 0 0 0 0 0 uid (relative to the solid) particle velocities, and p are the B C B C 0 0 0 0 0 0 0 0 0 bulk stress and uid pressure, respectively. Additionally, is B 3 C @ A 0 0 0 0 0 0 0 0 0 the uid density and = (1 ) + is the total (mixture) s f 0 0 0 0 0 0 0 0 0 density. Furthermore, the material parameters are de ned as 0 1 follows: B C 2 2 B C = K = K + M ; B C u u u m u 3 3 B C 1 1 B C M = BK = ; = 1 K =K (43e) B C u m u 2 B 1 2 C B C S = (46) B C 1 3 B w C B 2 C B C B C K 1=K 1=K m m s B C K = ; B = ; 0 B C 1 B 1=K 1=K + (1=K 1=K ) m s f s @ A (43f ) 0 0 where the subscript \u" denotes the quantities characterizing 0 0 0 0 1 2 with R = 1= 1= , K = K , and and = 1 2 0 2 1 2 the undrained solid matrix, while the subscript \m" denotes 0 0 =(c c ). The characteristic speeds of system (45) (the 2 i 1 2 the quantities characterizing the dry matrix. Thus, , K , u u eigenvalues of A) are the roots of the characteristic polynomial and are the undrained Lam e parameter, bulk modulus, and det(A I) = 0. These roots are given by the formulas the shear modulus of the undrained matrix, respectively, while K is the bulk modulus of the dry matrix. Additionally, K m s 2 2 = 0; = ; and K are the bulk moduli of grains and uid, respectively, while B is the so-called Biot parameter, and T is the tortuos- X + Y (X + Y ) 4(X Z )Y ity. Also, it is implied that the shear modulus of the undrained = ; (47a) matrix and that of the dry one coincide, = . u m X = K= + RK ; Y = =; Z = K=: (47b) Formula (47a) provides the so-called fast C and slow C 3 fast slow 4.2.2 Characteristic speeds characteristic speeds corresponding to longitudinal waves (fast and slow P-waves). Let us assume only elastic deformation of the porous material For Biot's model (43) we have and consider the linearized SHTC system (38) in one dimen- sion (x = x ) for the vector of state variables 0 1 0 0 z z Q = (v ; v ; v ; w ; w ; w ; p; s ; s ; s ): (44) 1 2 3 1 2 3 11 21 31 B C 0 0 B z z C A = ; z = T ; @ A M M 0 0 System (38) can be written in matrix form as follows: 2 M 0 0 u u Q + AQ = S; (45) (48) t x where and the characteristic polynomial is 4 2 A = det(A I) = + a + a = 0; (49a) 2 0 10 T ( + 2 ) + M ( 2 ) f u u f Parameters Biot SHTC Unit a = ; ( T ) Grain: f f M ( M 2 ) = 2500 2500 kg/m u u s 2 a = ; (49b) ( T ) f f C 4000 4332 m/s C - 3787 m/s sh while the characteristic speeds are given by the formula K = K = C 40 46.9 GPa s 2 2 p p 2 2 2 a + a 4a a a 4a = = C - 35.85 GPa 2 0 2 0 s 2 2 2 sh C = ; C = ; fast slow 2 2 Matrix, = 0:2: (49c) K 9:48 - GPa 24:5 - GPa which provides the fast and slow characteristic speeds corre- sponding to P-waves in Biot's model. T 3:75 - - 13 2 1 10 - m - 3:36 10 s Fluid: = 1040 1040 kg/m f 1 C 1500 1500 m/s K = K = C 2.34 2.34 GPa f 1 f 10 - Pa s Table 1: Material parameters for the Biot and SHTC models. Here C and C are the bulk sound speeds in the solid s f and uid, respectively. pure uid or pure solid cases. Figure 1: Comparison of the characteristic speeds for the Biot and SHTC models for the material parameters given in Table 1. Biot's model is taken in the version de- veloped by Carcione et al [6]. 4.2.3 Dispersion relations and sound speeds In Fig 1, for the two models, we compare the characteristic In this section, we will perform a plane wave analysis of the speeds C ( ) and C ( ) of the fast and slow modes. Some fast slow SHTC and Biot models. Moreover, we will consider only longi- other material parameters are given in Table 1. Note that, in tudinal waves. In this case the vector of SHTC state variables order to make the curves closer to each other, the bulk moduli reduces to of the solid phase in the two models are taken slightly di erent. In the limit cases ! 0 (pure solid) and ! 1 (pure Q = (v ; w ; p; s ): (50) 1 1 11 uid) the behavior of the characteristic speeds in the SHTC model better corresponds to that intuitively expected, i.e. the By linearizing the right hand-side of system (45) in a neigh- characteristic speeds of the mixture coincide with the char- borhood of Q = Q + q, where q is a small perturbation of acteristic speeds of the pure phases (horizontal dashed and an equilibrium state Q = (0; 0; p ; 0), we have 0 0 dashed-dotted lines), while the slow mode vanishes. In con- trast, in Biot's model the slow mode is present even in the q + Aq = Sq; (51) t x 11 where A and S = @S=@Q are matrices taken at the equilibrium nd that the dispersion relation is a bi-quadratic equation state Q : 4 2 + a + a = 0; (59a) 2 0 0 1 1 1 0 0 B C 0 0 R 0 B C A = ; !(T ( + 2 ) + M ( 2 )) @ A f u u f K K 0 0 a = ! M ( + 2 M ) u u 0 0 0 0 1 i( + 2 ) u u 0 0 0 0 + ; (59b) M ( + 2 M ) B C u u 0 0 0 B C S = : (52) @ A 0 0 0 0 0 0 0 0 !(T ) i f f a = ! : 2 2 M ( + 2 M ) M ( + 2 M ) u u u u We seek a solution that has the form (59c) i(!t kx) q = q ~e ; (53) The phase velocities V (!) and attenuation factor per ph which represents a plane harmonic wave of real frequency ! wavelength a (!) of the fast and slow sound waves for both and complex wave number k propagating in the direction x, models are shown in Fig. 2. Note that there is a di erence and q ~ = const is a constant vector of amplitudes. By substi- in the high-frequency limit of the fast modes, which is, in tuting (53) in (51), we arrive at a homogeneous linear system fact, due to the di erence between the characteristic speeds for q [24, 41]: C and C depicted in Fig. 1. To explain this, recall that fast slow the characteristic speeds (the eigenvalues of the homogeneous k i hyperbolic system) are the high-frequency limits (! ! 1) I A + S q = 0 (54) ! ! of the sound speeds (the eigenvalues of the non-homogeneous system (51)) [24, 41, 9]. The slow mode dispersion curves where I is the identity matrix of the same size as A and S. of both models are almost indistinguishable, see Fig. 2, the From (54) we have the following dispersion relation for (45): second column. Another conclusion is that there is a big di erence ( 10 ) k i det I A + S = 0: (55) in the attenuation factors a of the fast and slow modes for ! ! both models in the low frequency region, see Fig. 2, the second row. The phase velocity V (sound speed) and the attenuation ph factor a are then given by V = ; a = Im(k): (56) 5 Numerical test problems for ph Re(k) small-amplitude wave propagation In addition, it is convenient to use the attenuation per wave- length [41] 5.1 Finite di erence implementation 2V a Im(k) ph To discretize the governing equations, the velocity-stress for- a = a = = 2 ; (57) ! Re(k) mulation, which was proposed for elastic-wave equations in [20, 43], is used. The use of a numerical scheme on staggered where is the wavelength. grids is quite natural due to the fact that the equations form There are four solutions to (55) with matrices (52), which a symmetric rst order system of evolution equations for the are mixture components, relative velocities, pressure, and shear p stress. X+Y i(Y +Z)= 4Y (X Z)(i= 1)+(i(Y +Z)= (X+Y )) k(!)=! ; (58) 2(X Z)Y Following the notation of [43] and [16], we introduce a time- space grid with integer nodes t = nt, x = ix, y = jy i j where X; Y , and Z are de ned in (47), and = ! is the n+1=2 and half-integer nodes t = (n + 1=2)t, x = (i + i+1=2 non-dimensional frequency. 1=2)x, y = (j + 1=2)y , where t, x, and y denote j+1=2 Applying the same analysis to the Biot equations (43), we the grid step sizes for the temporal and spatial axes. 12 Figure 2: Phase velocities ( rst row) and attenuation factors per wavelength (second row) for the fast ( rst column) and slow (second column) modes of the SHTC and Biot models. The material parameters are given in Table 1. n n For a discrete function f = f (t ; x ; y ) de ned at the grid The wave eld components are de ned at di erent time- i j i;j nodes, let us introduce second-order centered nite di erence space grid nodes. The components of the mixture and n n operators: the relative velocities are de ned as (v ) , (v ) , x y i+1=2;j i;j+1=2 n n (w ) , (w ) , the pressure and the normal com- x y i+1=2;j i;j+1=2 n+1=2 n 1=2 (f ) (f ) n+1=2 n+1=2 i;j i;j ponents of the deviatoric stress as (p) , (s ) , D [f ] = ; xx i;j i;j i;j n+1=2 n+1=2 (s ) , and the shear stress as (s ) . n+1=2 n 1=2 yy xy i;j i+1=2;j+1=2 (f ) + (f ) i;j i;j A [f ] = ; (60) i;j To construct the nite di erence scheme, we use a nite volume approximation or balance law technique [42]. In this n n (f ) (f ) case the discrete form of equations (38) reads as i+1=2;j i 1=2;j D [f ] = ; i;j n 1=2 n 1=2 n n D [v ] = 1= D [P ] t x x (f ) (f ) i+1=2;j i+1=2;j i+1=2;j i;j+1=2 i;j 1=2 D [f ] = : (61) i;j n 1=2 n 1=2 0 0 + = D [s ] + D [s ] ; (62a) x xx y xy i+1=2;j i+1=2;j i+1=2;j The medium parameters are constant within each grid cell n 1=2 n 1=2 [x ; x ] [y ; y ], and they may have disconti- i 1=2 i+1=2 j 1=2 j+1=2 D [v ] = 1= D [P ] t y y i;j+1=2 i;j+1=2 i;j+1=2 nuities aligned with grid lines. This condition provides second- order convergence even for discontinuous medium parameters n 1=2 n 1=2 0 0 + = D [s ] + D [s ] ; (62b) x xy y yy i;j+1=2 i;j+1=2 i;j+1=2 [23]. n 1=2 n 1=2 0 0 discretization step: D [w ] = 1= 1= D [P ] t x x 1 2 i+1=2;j i+1=2;j i+1=2;j n 1=2 0 0 1 1 c c = A [w ] ; (62c) 2 t x 1 2 i+1=2;j i+1=2;j tC + 1: max 2 2 x y n 1=2 n 1=2 0 0 D [w ] = 1= 1= D [P ] A forcing function f (t; x; y) is introduced as the source term t y y 1 2 i;j+1=2 i;j+1=2 i;j+1=2 in the right-hand side of the pressure equation or the equations n 1=2 0 0 for the normal components of the deviatoric stress in system c c = A [w ] ; (62d) 2 t y 1 2 i;j+1=2 i;j+1=2 (38). In both cases, a volumetric-type source term is obtained. The source function is de ned as the product of Dirac's delta function in space and Ricker's wavelet in time: 4 2 n n n D [s ] = () D [v ] D [v ] t xx i;j x x y y i;j i;j i;j 3 3 2 2 2 2 2 2 f (t) = (1 2 f (t t ) )exp[ f (t t ) ]; (64) 0 0 0 0 0 n ( = ) A [s ] ; (62e) i;j t xx 2 i;j where f is the source peak frequency and t is the wavelet 0 0 delay. 4 2 No special care is taken to suppress the outgoing waves n n n D [s ] = () D [v ] D [v ] t yy i;j y y x x i;j i;j i;j 3 3 with the help of absorbing boundary conditions (for exam- ple, PML). The simulation is stopped before the waves have 0 n ( = ) A [s ] ; (62f ) i;j t yy 2 i;j reached the boundaries of the computational domain. The nu- merical experiments have been performed on a desktop com- puter with Intel(R) Core(TM) i7 3.60 GHz processor. In the subsequent sections, we will illustrate the main fea- n n D [s ] = fg D [v ] t xy x y i+1=2;j+1=2 i+1=2;j+1=2 i+1=2;j+1=2 tures of wave eld formation and propagation depending on n 0 n porosity , friction parameter , source peak frequency f , +D [v ] = A [s ] ; 2 0 y x t xy i+1=2;j+1=2 2 i+1=2;j+1=2 i+1=2;j+1=2 and shear relaxation time . Before doing this, we start with (62g) considering the homogeneous dissipation-less case. n n n D [P ] = (K ) D [v ] + D [v ] t i;j x x y y i;j i;j i;j 5.2 Dependence on porosity . 0 0 0 0 0 n n ( ) K= D [w ] + D [w ] ; (62h) x x y y 2 1 1 2 i;j i;j i;j The computational domain = [ 0:65; 0:65] m was dis- where the e ective medium parameters on the staggered grids cretized with N N grid points, N = N = 3250, which x y x y are obtained as volume arithmetic or harmonic averaging [23]: amounts to 10 points per slow compressional wavelength in for a source of central frequency f = 10 Hz and for various hfi = (f + f )=2; i;j i+1;j values of porosity . The model parameters were taken from i+1=2;j Table 1. The source was located in the center of the computa- hfi = (f + f )=2; i;j i;j+1 i;j+1=2 tional domain. The propagation time was chosen to be equal 1 1 1 1 4 4 to T = 1:1 10 s with source time delay t = 1=f =0:1 10 0 0 0 ffg = 4= + + + : i+1=2;j+1=2 f f f f i;j i+1;j i;j+1 i+1;j+1 s. The time step was chosen according to the classical Courant (63) stability criterion for staggered grids with CFL = 0:9. Fig. 3(a) shows a snapshot of the mixture velocity v at time T , on the whole computational domain. The porosity parameter was chosen to be equal to =0, which corresponds The thus obtained 2D velocity-stress nite di erence scheme to the case of a pure elastic solid. In an elastic medium, only is second-order accurate in both time and space. Note that the one fast P-wave with a velocity equal to 6155 m/s is excited approximation can be easily improved by using higher order from a source of volumetric type. The seismogram in Fig. operators. The well-known Courant-Friedrichs-Lewy (CFL) 3(b) con rms the occurrence of this wave with the predicted stability criterion also holds for this case: the time step must velocity. The receivers were located along the x-axis starting be chosen small enough in order that the fastest characteristic from the source point towards the boundary with a uniform wave (P-wave) C travel a distance smaller than the spatial spacing between the receivers. max 14 -0.6 -0.1 -0.4 -0.2 0.1 0.2 0.3 0.2 0.4 0.5 0.4 0.6 0.6 2 4 6 8 10 -5 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 Time (s) X(m) (b) (a) Figure 3: Wave eld in the pure elastic medium generated by the Ricker wavelet of a volumetric type. A snapshot at time 4 1 5 T = 1:1 10 s (a) and seismogram (b) of the horizontal mixture velocity v for source of central frequency f = 10 0 0 Hz. Fig. 4 shows the results of calculations similar to the previ- model, can be viewed as a friction parameter, because it ous ones but with porosity parameter =1. This corresponds controls interfacial friction in a multiphase medium and leads to the case of a pure liquid with one pressure wave with a ve- to wave dispersion and attenuation. locity of 1500 m/s. The numerical propagation velocity can be Consider the same homogeneous numerical model as in the easily estimated from the computed seismogram in Fig. 4(b) previous sections with equal to 3:36 10 from Table 1. to be exactly 1500 m/s. Let us observe the behavior of the P-waves if we increase or The simulation results for porosity parameter =0.5 are decrease this parameter twice. Fig. 8(a-c) shows snapshots of shown in Fig. 5. The fast P-wave velocity is estimated to be the mixture velocity v for these three values of . Signif- 4100 m/s from the computed seismogram in Fig. 5(b), which icant wave amplitude variations are observed only for slow is consistent with the data from Table 1. In the snapshot of P-wave. More detailed variations of the amplitudes can be Fig. 5(a) we do not observe the predicted slow P-wave because seen in Fig. 9, where the wave eld along the x-axis at y = 0 it is completely attenuated, not visible at this time, and its is presented. This gure demonstrates an increase in the am- amplitude is very small. However, if we zoom in the image, plitude of the slow P-wave with increasing , and vice versa. we will be able to see this slow P-wave in Fig. 6. The ampli- A change in the form of slow waves is also observed, while the tude variation for several values of can also be seen in Fig. 7, fast wave remains almost unchanged. where the wave eld distribution along the x-axis for y = 0 is Summarizing the results of numerical experiments of this presented. section, we conclude that by varying the parameter in sys- Summarizing the results of numerical experiments of this tem (38) it is possible to a ect the amplitude and propagation section, we conclude that by varying porosity in system velocity of the slow P-wave. One of the interesting applica- (38), it is possible to correctly describe the three states of the tions, in our opinion, can be the solution of the inverse problem medium: liquid, solid, and poroelastic. of determining the coecient by analyzing the ratio of the amplitudes of the fast and slow waves in a eld experiment. 5.3 Dependence on friction . 5.4 Dependence on frequency f . In this section, we study the behavior and properties of the fast and slow P-waves, depending on the parameter . This In this section, we perform a study of dispersion of the wave parameter is present as the denominator in the right-hand side velocity depending on the source peak frequency f based on of the second equation in system (38). By analogy with Biot's the same homogeneous numerical model with parameters from Y(m) Receiver x coordinate (m) -0.6 -0.2 -0.1 -0.4 -0.2 0.1 0.2 0.3 0.2 0.4 0.4 0.5 0.6 0.6 2 4 6 8 10 -5 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 Time (s) X(m) (b) (a) Figure 4: Wave eld in the pure uid generated by the Ricker wavelet of a volumetric type. A snapshot at time T = 1:1 10 1 5 s (a) and seismogram (b) of the mixture horizontal velocity v for source of central frequency f = 10 Hz. -0.6 -0.2 -0.4 -0.1 -0.2 0.1 0.2 0.3 0.2 0.4 0.4 0.5 0.6 0.6 2 4 6 8 10 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 -5 Time (s) X(m) (b) (a) Figure 5: Wave eld in the porous medium with porosity = 0:5 generated by the Ricker wavelet of a volumetric type. A 4 1 snapshot at time T = 1:1 10 s (a) and seismogram (b) of the mixture horizontal velocity v for source of central frequency f = 10 Hz. Y(m) Y(m) Receiver x coordinate (m) Receiver x coordinate (m) -0.1 -0.02 -0.05 0.02 0.04 0.06 0.05 0.08 0.1 0.1 0.12 0.5 1 1.5 2 2.5 3 -5 Time (s) -0.1 -0.05 0 0.05 0.1 X(m) (b) (a) Figure 6: Wave eld in the porous medium with porosity = 0:5 generated by the Ricker wavelet of a volumetric type. A 5 1 zoom of snapshot at time 1 10 s (a) and seismogram (b) of the mixture horizontal velocity v for source of central frequency f = 10 Hz. 0.01 =0.5 =0.8 0.008 =0.2 0.006 0.004 0.002 -0.002 -0.004 -0.006 -0.008 -0.01 -0.13 -0.065 0 0.065 0.13 X (m) Figure 7: The distribution of the mixture horizontal velocity v generated by the Ricker wavelet of a volumetric type along the x-axis at time 2:1 10 s and y = 0 for several values of porosity : = 0:2 (black), = 0:5 (blue), = 0:8 (red). Amplitude (m) Y(m) Receiver x coordinate (m) -0.1 -0.1 -0.1 -0.05 -0.05 -0.05 0 0 0 0.05 0.05 0.05 0.1 0.1 0.1 -0.1 -0.05 0 0.05 0.1 -0.1 -0.05 0 0.05 0.1 -0.1 -0.05 0 0.05 0.1 X(m) X(m) X(m) 7 7 7 (a) = 3:36 10 (b) = 2 3:36 10 (c) = 1=2 3:36 10 2 2 2 Figure 8: Wave eld in the porous medium with porosity = 0:5 generated by the Ricker wavelet of a volumetric type. Snapshots 5 1 at time 2:1 10 s of the horizontal mixture velocity v for several values of . -3 -1 -2 -3 -4 -0.13 -0.065 0 0.065 0.13 X (m) 5 1 7 Figure 9: A comparison at time 2:1 10 s of values of the mixture velocity v for several values of : = 3:36 10 (blue), 2 2 7 7 = 2 3:36 10 (red), = 1=2 3:36 10 (black). 2 2 Y(m) Amplitude (m) Y(m) Y(m) Table 1 as in the previous sections. The dispersion curves in 5.6 Layered media. Fig. 1 show that the main velocity changes are in the range of 4 6 This test illustrates the e ects of the interfaces between pure 10 10 Hz. Because of a di erence of three orders of mag- uid, poroelastic and pure elastic media on an example of a nitude in the frequency range, our comparison will be made three-layered medium with the media parameters from Table not for a single computational domain, but for three di erent 1. The size of the computational domain is 0:025 m in the x domains. More speci cally, we use a ten orders of magnitude and y directions. The upper layer is water, the lower layer is scaling of both space and time. Fig. 10 presents snapshots, 1 an elastic medium and in the middle, from 0:005 m to 0:005 and Fig. 11 presents seismograms of the mixture velocity v m, there is a poroelastic layer with porosity =0.2. A source for several frequencies. For each time frequency, a snapshot is of central frequency f = 10 Hz is located in the water layer recorded at time 5=f (including a shift wavelet delay of 1=f ) 0 0 4 at the point x = 0; y = 0:01. for square domains with a side of 5 m (for f = 10 ), 0:5 m (for 5 6 In order to present all types of waves, let us consider snap- f = 10 ), and 0:05 m (for f = 10 ). The time and size are 0 0 shots of the total velocity vector for di erent moments of time. chosen in such a way that in the isotropic elastic case we can We strongly ampli ed the wave eld amplitude in Fig. 13 to be obtain three identical snapshots. As expected in the poroe- able to pick out the slow compressional waves in the snap- lastic case, we observe wave eld di erences which are most shots. In order to interpret the waves arising in the medium, clearly seen in the seismograms: the lower the frequency, the we use 'P' to mark P-waves and 'S' to mark S-waves. The stronger the dispersion and attenuation of the slow P-wave. subscript 'r' indicates the re ected waves, while the subscript To estimate the phase velocity, we use a spectral ratio tech- 't' indicates the transmitted waves. Also, we use the subscript nique [17], [7] and obtain a phase velocity v = 420 m/s for 5 's' to identify a slow P-wave and the subscript 'f ' to identify a frequency f = 10 Hz and v = 670 m/s for the frequency 0 p 6 a pressure wave in the uid layer. f = 10 Hz. It is not possible to make similar estimation for The source in the water layer excites a pressure wave ( de- a frequency f = 10 due to the fact that the slow wave has a noted by P in Fig. 13) which propagates towards the poroe- low amplitude. We can also conclude that the attenuation of lastic layer. This wave then re ects from the bottom of the the slow P-wave is suciently strong even for high frequencies. water-poroelastic interface (P P ) and generates a fast trans- f r mitted P-wave (P P ), a slow transmitted P-wave (P P ), and f t f s a transmitted S-wave (P S ) in the poroelastic medium. Af- f t terwards, these waves generate a family of transmitted and 5.5 Dependence on shear relaxation time. re ected waves (including a transmitted P-wave (P P P ), an f t t S-wave (P S S ), and a re ected slow P-wave (P P P )) from f t t f t s The aim of this section is to show that there is an additional the upper and lower boundaries of the poroelastic layer. mechanism of energy dissipation embedded in system (38) and In this example we see that the propagation of all types of controlled by the relaxation parameter in the right-hand side waves predicted by the elasticity and poroelasticity theories of equation (38d). A proper choice of the relaxation time is correct: slow waves arise only in the poroelastic layer, only allows one to model irreversible (elastoplastic) deformations pressure waves propagate in the liquid layer, while longitudinal in the solid matrix, e.g. [8, 33], or viscous ows [9]. and shear waves appear in the elastic medium. All the previous numerical examples were simulated without relaxation of tangential shear stresses, that is, the right-hand side in (38d) vanished. Formally, this corresponds to the case 6 Conclusions of = 1. It is quite obvious that, for nite values of , the mechanism of relaxation of tangential stresses provides an An extension of the uni ed model of continuum uid and additional ability of the model to control wave attenuation. solid mechanics [9] for compressible uid ows in elastoplas- To this end, we again consider the example from the previous tic porous media has been proposed. The derivation is based Section 5.4 for a frequency f = 10 Hz and compare the so- on the Symmetric Hyperbolic Thermodynamically Compati- lution with a similar test, but with allowance for relaxation of ble (SHTC) theory [32], and the resulting model represents a tangential stresses. As expected, a comparison of the seismo- combination of the uni ed continuum model from [9] with the grams in Fig. 12 shows that relaxation of tangential stresses SHTC model for two-phase compressible ows from [36]. The leads to dispersion and attenuation of seismic waves. On the governing equations satisfy two laws of thermodynamics (en- distant receivers, it is seen that the smaller , the greater the ergy conservation and non-decreasing of entropy) and form a attenuation. On the nearer receivers, dispersion probably has rst-order symmetric system which is hyperbolic in the sense a predominant e ect. of Friedrichs [11] if the generating thermodynamic potential is 19 -2.5 -0.25 -0.025 -2 -0.2 -0.02 -1.5 -0.15 -0.015 -1 -0.1 -0.01 -0.5 -0.05 -0.005 0 0 0 0.5 0.05 0.005 0.1 0.01 0.015 0.15 1.5 0.02 0.2 0.025 0.25 2.5 -0.02 -0.01 0 0.01 0.02 -0.2 -0.1 0 0.1 0.2 -2 -1 0 1 2 X(m) X(m) X(m) 4 (c) f = 10 (b) f = 10 (a) f = 10 0 Figure 10: Wave eld in the porous medium with porosity = 0:5 generated by the Ricker wavelet of a volumetric type. 1 4 5 6 Snapshots of the horizontal mixture velocity v for several frequencies: f = 10 Hz (a), f = 10 Hz (b), f = 10 0 0 0 Hz (c). -0.5 -0.05 -3 -5 0 0 0.5 0.05 1 0.1 1.5 0.15 2 0.2 20 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 -4 -5 -6 Time (s) Time (s) Time (s) 10 10 10 4 5 6 (a) f = 10 (b) f = 10 (c) f = 10 0 0 0 Figure 11: Wave eld in the porous medium with porosity = 0:5 generated by the Ricker wavelet of a volumetric type. 1 4 5 Seismograms of the horizontal mixture velocity v for several frequencies: f = 10 Hz (a), f = 10 Hz (b), 0 0 f = 10 Hz(c). Receiver x coordinate (m) Y(m) Receiver x coordinate (m) Y(m) Receiver x coordinate (m) Y(m) -0.02 0.02 0.04 0.06 0.08 0.1 0.12 0.5 1 1.5 2 2.5 3 -4 Time (s) Figure 12: Comparison of seismograms of the horizontal mixture velocity v for several values of shear relaxation time : no 5 4 relaxation (black), = 10 (red), = 10 (blue). convex. tive agreement. A number of two-dimensional test problems has been solved Based on the above-proposed non-linear model, a linearized for the propagation of small-amplitude waves described by the rst order PDE system for small-amplitude wave propagation formulated model. These test problems include, in particular, in a stationary saturated porous medium has been derived. the study of the dependence of wave elds on model parame- The linear system is written in terms of the velocity of the ters. The numerical results demonstrate that the SHTC model solid- uid mixture, and the relative velocity of the phase mo- describes correctly all physical characteristics of the process. tion, pressure and shear stress. Such a formulation allowed us Finally, we note that the developed poroelastic model is a straightforward development of an ecient nite di erence based on the SHTC formulation of mixtures. It thus should scheme on a staggered grid. be kept in mind that other approaches to obtain continuum A comparison of the above-proposed SHTC model and the models for mixtures are possible [4, 26, 27]. In particu- classical Biot model for wave propagation in a saturated elas- lar, on the PDE level, the continuum formulation for mix- tic porous medium has been made at the aid of the dispersion tures obtained in the GENERIC (General Equation for Non- analysis. It turns out that, although the basic equations of the Equilibrium Reversible-Irreversible Coupling) framework [26, two models are di erent, the SHTC model is able to describe 28] di ers from the SHTC formulation for mixtures used in all the e ects (in particular, the existence of a slow P-wave) k l @w @w this paper by a term E l missing in equation predicted by Biot's theory, in good quantitative and qualita- @x @x l k Receiver x coordinate (m) (14e). Also, on the physical level, the GENERIC formulation [2] M. Biot. \Theory of propagation of elastic waves in uid-saturated porous solid. I. Low-frequency range". Journal of the Acoustical may di er by a di erent interpretation of the state variables, Society of America 28.2 (1956), pp. 168{178 (Cited on p. 1). e.g. see a discussion on the SHTC compatible and alternative [3] M. Biot. \Mechanics of deformation and acoustic propagation in Poisson brackets for heat conduction and the total momentum porous media:" Journal of Applied Physics 33.4 (1962), pp. 1482{ de nition in [32]. Nevertheless, for the case of small amplitude 1498 (Cited on p. 1). waves, such di erences should not result in a di erent linear [4] A. M. Blokhin and V. N. Dorovsky. Mathematical Modelling in the system and, in particular, the extra term mentioned above Theory of Multivelocity Continuum. Nova Science Publishers Inc., k l @w @w New York, 1995, p. 183 (Cited on pp. 1, 21). should vanish similar to the term v , see equa- @x @x l k [5] S. Busto, S. Chiocchetti, M. Dumbser, E. Gaburro, and I. Peshkov. tion (38b). Perhaps, di erences in solutions should appear in \High Order ADER Schemes for Continuum Mechanics". Frontiers case of nite deformations of poroelastic continuum. We hope in Physics 8 (2020). doi: 10.3389/fphy.2020.00032 (Cited on to investigate this in detail in future publications. p. 3). [6] J. M. Carcione, C. Morency, and J. E. Santos. \Computational poroelasticity | A review". Geophysics 75.5 (2010), 75A229{ Acknowledgements 75A243. doi: 10.1190/1.3474602 (Cited on pp. 1, 9, 11). [7] E. Caspari, M. Novikov, V. Lisitsa, N. D. Barbosa, B. Quintal, J. G. The authors are grateful to M. Yudin for valuable help in Rubino, and K. Holliger. \Attenuation mechanisms in fractured uid-saturated porous rocks: a numerical modelling study". Geo- the manuscript preparation. The research of E.R. and G.R. physical Prospecting 67.4 (2019), pp. 935{955. doi: 10.1111/1365- in Sects.2-4 was supported by the Russian Science Foundation 2478.12667 (Cited on p. 19). under grant 19-77-20004, the research in Sect.5 was supported [8] M. Dumbser, I. Peshkov, and E. Romenski. \A Uni ed Hyper- by the Russian Foundation for Basic Research under grant 19- bolic Formulation for Viscous Fluids and Elastoplastic Solids". 01-00347. I.P. gratefully acknowledges the support of Agence In: Theory, Numerics and Applications of Hyperbolic Problems II. HYP 2016. Ed. by C. Klingenberg and M. Westdickenberg. Nationale de la Recherche (FR) (grant ANR-11-LABX-0040- Vol. 237. Springer Proceedings in Mathematics and Statistics. CIMI) under program ANR-11-IDEX-0002-02. The work Cham: Springer International Publishing, 2018, pp. 451{463. doi: of M.D. and I.P. was partially supported by the European 10.1007/978-3-319-91548-7_34 (Cited on pp. 3, 9, 19). Union's Horizon 2020 Research and Innovation Programme [9] M. Dumbser, I. Peshkov, E. Romenski, and O. Zanotti. \High order under project ExaHyPE. M.D. and I.P. also gratefully ac- ADER schemes for a uni ed rst order hyperbolic formulation of knowledge funding from the Italian Ministry of Education, continuum mechanics: Viscous heat-conducting uids and elastic solids". Journal of Computational Physics 314 (2016), pp. 824{ University and Research (MIUR) under the Departments of 862. doi: 10.1016/j.jcp.2016.02.015 (Cited on pp. 2, 3, 9, 12, Excellence Initiative 2018{2022 attributed to DICAM of the 19). University of Trento, as well as nancial support from the [10] M. Dumbser, I. Peshkov, E. Romenski, and O. Zanotti. \High order University of Trento under the Strategic Initiative Modeling ADER schemes for a uni ed rst order hyperbolic formulation of and Simulation. I.P. has further received funding from the Newtonian continuum mechanics coupled with electro-dynamics". Journal of Computational Physics 348 (2017), pp. 298{342. doi: University of Trento via the UniTN Starting Grant initiative. 10.1016/j.jcp.2017.07.020 (Cited on p. 2). [11] K. O. Friedrichs. \Symmetric positive linear di erential equa- tions". Communications on Pure and Applied Mathematics 11.3 In memoriam (1958), pp. 333{418. doi: 10 . 1002 / cpa . 3160110306 (Cited on pp. 2, 19). This paper is dedicated to the memory of Dr. Douglas Nelson th th [12] S. K. Godunov. \An interesting class of quasilinear systems". Dokl. Woods ( January 11 1985 - ySeptember 11 2019), promis- Akad. Nauk SSSR 139(3) (1961), pp. 521{523 (Cited on p. 2). ing young scientist and post-doctoral research fellow at Los [13] S. K. Godunov, T. Y. Mikha^ lova, and E. I. Romenski^ . \Systems Alamos National Laboratory. Our thoughts and wishes go to of thermodynamically coordinated laws of conservation invari- his wife Jessica, to his parents Susan and Tom, to his sister ant under rotations". Siberian Mathematical Journal 37.4 (1996), Rebecca and to his brother Chris, whom he left behind. pp. 690{705. doi: 10.1007/BF02104662 (Cited on p. 2). [14] S. K. Godunov and E. I. Romenski. \Thermodynamics, conser- vation laws, and symmetric forms of di erential equations in me- References chanics of continuous media". In: Computational Fluid Dynamics Review 95. John Wiley, NY, 1995, pp. 19{31 (Cited on p. 2). [1] M. Biot. \Theory of propagation of elastic waves in a uid- [15] S. K. Godunov and E. Romenskii. Elements of Continuum Me- saturated porous solid. II. Higher frequency range". Journal of the chanics and Conservation Laws. Springer US, 2003, p. 258 (Cited Acoustical Society of America 28.2 (1956), pp. 179{191 (Cited on on pp. 2, 3). p. 1). 22 [16] R. W. Graves. \Simulating seismic wave propagation in 3D elas- [31] I. Peshkov and E. Romenski. \A hyperbolic model for viscous New- tic media using staggered-grid nite di erences". Bulletin of the tonian ows". Continuum Mechanics and Thermodynamics 28.1-2 Seismological Society of America 86.4 (Aug. 1996), pp. 1091{1106 (2016), pp. 85{104. doi: 10.1007/s00161-014-0401-6 (Cited on (Cited on p. 12). pp. 2, 9). [17] B. Gurevich and R. Pevzner. \How frequency dependency of Q [32] I. Peshkov, M. Pavelka, E. Romenski, and M. Grmela. \Continuum a ects spectral ratio estimates". GEOPHYSICS 80.2 (2015), A39{ mechanics and thermodynamics in the Hamilton and the Godunov- A44. doi: 10.1190/geo2014-0418.1 (Cited on p. 19). type formulations". Continuum Mechanics and Thermodynamics 30.6 (2018), pp. 1343{1378. doi: 10.1007/s00161- 018- 0621- 2 [18] H. Jackson and N. Nikiforakis. \A numerical scheme for non- (Cited on pp. 2, 3, 4, 19, 22). Newtonian uids and plastic solids under the GPR model". Journal of Computational Physics 387 (2019), pp. 410{429. doi: 10.1016/ [33] I. Peshkov, W. Boscheri, R. Loub ere, E. Romenski, and M. Dumb- j.jcp.2019.02.025 (Cited on p. 3). ser. \Theoretical and numerical comparison of hyperelastic and hypoelastic formulations for Eulerian non-linear elastoplasticity". [19] A. Khoei and T. Mohammadnejad. \Numerical modeling of multi- Journal of Computational Physics 387 (2019), pp. 481{521. doi: phase uid ow in deforming porous media: A comparison between 10.1016/j.jcp.2019.02.039 (Cited on pp. 3, 19). two- and three-phase models for seismic analysis of earth and rock- ll dams". Computers and Geotechnics 38.2 (2011), pp. 142{166. [34] E. Rohan and V. Luke s. \Modeling large-deforming uid-saturated doi: 10.1016/j.compgeo.2010.10.010 (Cited on p. 1). porous media using an Eulerian incremental formulation". Ad- vances in Engineering Software 113 (2017), pp. 84{95. doi: 10. [20] A. R. Levander. \Fourth-order nite-di erence P-SV seismo- 1016/j.advengsoft.2016.11.003 (Cited on p. 1). grams". Geophysics 53.11 (Nov. 1988), pp. 1425{1436. doi: 10. 1190/1.1442422 (Cited on p. 12). [35] E. Romenski. \Conservative formulation for compressible uid ow through elastic porous media". In: Vzquez-Cendn et al. (eds) [21] Y. J. Masson, S. R. Pride, and K. T. Nihei. \Finite di erence Numerical Methods for Hyperbolic Equations. Taylor & Francis modeling of Biot's poroelastic equations at seismic frequencies". Group, London (2013), pp. 193{200 (Cited on pp. 2, 5). Journal of Geophysical Research: Solid Earth 111,B10305 (2006). doi: 10.1029/2006JB004366 (Cited on pp. 1, 9). [36] E Romenski, D Drikakis, and E Toro. \Conservative Models and Numerical Methods for Compressible Two-Phase Flow". Journal [22] A. Merxhani. \An introduction to linear poroelasticity" (2016), of Scienti c Computing 42(1) (2010), pp. 68{95 (Cited on pp. 2, pp. 1{38 (Cited on p. 9). 3, 4, 6, 19). [23] P. Moczo, J. Kristek, V. Vavrycuk, R. J. Archuleta, and L. Ha- [37] E Romenski, A. D. Resnyansky, and E. F. Toro. \Conservative lada. \3D heterogeneous staggered-grid nite-di erence modeling hyperbolic model for compressible two-phase ow with di erent of seismic motion with volume harmonic and arithmetic averaging phase pressures and temperatures". Quarterly of applied mathe- of elastic moduli and densities". Bull. Seism. Soc. Am. 92.8 (2002), matics 65(2).2 (2007), pp. 259{279 (Cited on p. 2). pp. 3042{3066 (Cited on pp. 13, 14). [38] E. Romenski. \Hyperbolic systems of thermodynamically compati- [24] A Muracchini, T Ruggeri, and L Seccia. \Dispersion relation in the ble conservation laws in continuum mechanics". Mathematical and high frequency limit and non linear wave stability for hyperbolic computer modelling 28(10) (1998), pp. 115{130 (Cited on p. 2). dissipative systems". Wave Motion 15.2 (1992), pp. 143{158 (Cited on p. 12). [39] E. Romenski, A. A. Belozerov, and I. M. Peshkov. \Conservative formulation for compressible multiphase ows". Quarterly of Ap- [25] S. Ndanou, N. Favrie, and S. Gavrilyuk. \Criterion of Hyperbolicity plied Mathematics 74.1 (2016), pp. 113{136. doi: 10.1090/qam/ in Hyperelasticity in the Case of the Stored Energy in Separable 1409 (Cited on pp. 2, 3, 4, 6). Form". Journal of Elasticity 115.1 (2014), pp. 1{25. doi: 10.1007/ s10659-013-9440-7 (Cited on p. 3). [40] E. Romensky. \Thermodynamics and Hyperbolic Systems of Bal- ance Laws in Continuum Mechanics". In: Toro E.F. (eds) Go- [26] M. Pavelka, V. Klika, and M. Grmela. Multiscale Thermo- dunov Methods. Springer, Boston, MA (2001), pp. 745{761 (Cited Dynamics. Berlin, Boston: De Gruyter, 2018. doi: 10 . 1515 / on p. 2). 9783110350951 (Cited on p. 21). [41] T. Ruggeri and M. Sugiyama. Rational Extended Thermodynamics [27] M. Pavelka, F. Mar s k, and V. Klika. \Consistent theory of mix- beyond the Monatomic Gas. Cham: Springer International Publish- tures on di erent levels of description". International Journal of ing, 2015, pp. 1{376. doi: 10.1007/978-3-319-13341-6 (Cited on Engineering Science 78 (2014), pp. 192{217. doi: 10 . 1016 / j . p. 12). ijengsci.2014.02.003 (Cited on p. 21). [42] A. A. Samarskii. The Theory of Di erence Schemes. CRC Press, [28] M. Pavelka, V. Klika, O. Esen, and M. Grmela. \A hierarchy of 2001, p. 786 (Cited on p. 13). Poisson brackets in non-equilibrium thermodynamics". Physica D: Nonlinear Phenomena 335 (2016), pp. 54{69. doi: 10.1016/j. [43] J. Virieux. \P-SV wave propagation in heterogeneous media; physd.2016.06.011 (Cited on p. 21). velocity-stress nite-di erence method". Geophysics 51.4 (Apr. 1986), pp. 889{901. doi: 10.1190/1.1442147 (Cited on p. 12). [29] F. Pesavento, B. A. Schre er, and G. Scium e. \Multiphase Flow in Deforming Porous Media: A Review". Archives of Computational [44] K. Wilmanski. \A few remarks on Biot's model and linear acoustics Methods in Engineering 24.2 (2017), pp. 423{448. doi: 10.1007/ of poroelastic saturated materials". Soil Dynamics and Earthquake s11831-016-9171-6 (Cited on p. 1). Engineering 26.6-7 (2006), pp. 509{536. doi: 10.1016/j.soildyn. 2006.01.006 (Cited on pp. 1, 8). [30] I. Peshkov, M. Grmela, and E. Romenski. \Irreversible mechanics and thermodynamics of two-phase continua experiencing stress- [45] K. Wilmanski. \A Thermodynamic Model of Compressible Porous induced solid- uid transitions". Continuum Mechanics and Ther- Materials with the Balance Equation of Porosity". Transport modynamics 27.6 (2015), pp. 905{940. doi: 10.1007/s00161-014- in Porous Media 32.1 (1998), pp. 21{47. doi: 10 . 1023 / A : 0386-1 (Cited on pp. 2, 4). 1006563932061 (Cited on p. 1). 23 [46] K. W. Winkler, H.-L. Liu, and D. L. Johnson. \Permeability and borehole Stoneley waves: Comparison between experiment and the- ory". GEOPHYSICS 54.1 (1989), pp. 66{75. doi: 10 . 1190 / 1 . 1442578 (Cited on p. 1). 24 2 Figure 13: Snapshots of the norm of mixture velocity jjvjj for a layered medium: water (upper layer, = 1), poroelastic (middle layer, = 0:2) and elastic solid (bottom layer, = 0).
Journal
Physics – arXiv (Cornell University)
Published: Oct 9, 2019