Mesoscale structure formation in high-speed deformation of solids
Khantuleva T.A., Klimanek P., Mescheryakov
Yu.I.
St.-Petersburg State University RUSSIA
Tech. University Freiberg Institute of Metallurgy GERMANY
Institute of Mech. Eng. St.-Petersburg , RUSSIA
Introduction
Microstructure evolution in plastic deformation
and its influence on the mechanical properties of solids is now as before an
actual topic of materials science and technology. At the present time
the interest is directed
in particular to the structure property interrelations occurring
in deformation at high temperatures, high strain rates and/or large strains.
Processes of this kind in crystalline materials are characterized by
i)
high content
and complicated arrangements of lattice defects (e.g. dislocation densities
> 1014 m-2 , formation of dense dislocation walls,
subboundaries, microbands etc.)
ii)
microstructure evolution on different length scales (Microscale:
slip of individual dislocations, stacking- fault formation, microtwinning; Mesoscale:
formation of dislocation walls and subboundaries accompanied by the
generation of local lattice disorientations (microtextures), nucleation and
growth of new grains by dynamic or postdynamic recrystallisation), and
iii)
formation of significant local
inhomogeneity and/or instability of the deformation documented by strain
gradients, shear banding, local cracking etc.
The present paper is concerned with the
influence of the strain rate on the microstructure formation in metallic
materials at temperatures
T<<Tm (Tm – melting temperature). In
this connection two limiting cases of the deformation behavior have to be
distinguished:
· At small strain rates
and applied
stresses 
(
- yield stress) the deformation of a crystalline solids is
elastic- plastic and macroscopically described by classical continuum
mechanics. Its elementary mechanisms are shear processes related to the lattice
structure (i.e. dislocation slip, mechanical twinning, or strain-induced
martensite formation ), and the evolution of the
microstructure can be explained in wide limits in terms of arrangements of
lattice defects, e.g. on the base of dislocation-disclination dynamics ( ).
· At high strain rates 
and applied stresses 
(i.e. on the conditions of shock loading) the shear resistance (and,
consequently, the lattice structure) of the solid becomes practically
negligible amd the deformation can be treated, in principle, within the
framework of hydrodynamics ( Navier-Stokes equations) of incompressible viscous
fluids ( ). However, as Navier-Stokes equations
describe the behavior of a structureless media the treatment cannot explain the
formation of experimentally observed (…) non-crystallographic structure
defects.
In
between the limits nonequilibrium transport of mass, momentum and energy is
followed by the self-organization at the intermediate scale level between
macro- and micro-levels. But till present there was no fundamental theory to
describe structure formation in this the most general case as far as the
conception of nonequilibrium transport itself had not been developed.
Nonequilibrium transport
processes
Transport of mass, momentum and energy follows
all real physical processes. For distributed systems mass 
, momentum p (p ∕=v), and internal
energy E densities are used to
describe the transport processes in media. The densities should satisfy the
general balance equations resulted from the basic nature law confirming the
full energy conservation.
J , Q are the momentum and energy fluxes and 
are the momentum and
energy sources. If the fluxes are expressed through the densities and the
sources are given, the balance equations govern the mass, momentum and internal
energy distributions or so called macroscopic fields. However, the problem of
the determining relationships between the fluxes and densities is solved only
in two limiting cases: for small velocity gradients or strain-rates (classical
hydrodynamics) and small deformations (linear theory of elasticity). In both
limits the balance equations are the partial differential ones derived under
the conventional condition 
. It means that the typical linear size of the internal
structure element l can be neglected
compared to the typical scale of the velocity field inhomogeneity and the
medium considers being structureless as the continuum mechanics supposes. For
high-rates and large deformations a medium demonstrates the internal structure
effects, and the problem of the determining relationships is still exists as
far as this processes can be referred to highly nonequilibrium ones.
On the condition 
a new problem arises,
how to determine the concept of the macroscopic density in nonequilibrium. The
classical definition ceases its conventional meaning and it is necessary to
involve the probability concepts and mathematical statistics. In scope of the
statistical theory the averaged values are defined independently on the scale
parameters in terms of statistical distributions characterized by mathematical
expectation and dispersion. In the case where the statistical distribution is
close to the normal Gaussian distribution in medium element velocities it
determines the macroscopic densities as the first statistical moments. The law
of large numbers supposes the conventional scale condition satisfied. This case
corresponds to the thermodynamic equilibrium where at microscale chaotic heat
fluctuations take place. Chaos means an independence of fluctuations in
different points of a system and in different time instances. So, in
equilibrium all events at microscale are considered as being independent.
It had been experimentally found out
that as the strain-rates growing a synchronization of the fluctuations begins.
The fluctuations cease being independent and there appears a correlation
between states of the system in different space regions and in different time
intervals. In the nonequilibrium mathematical statistics there exists a concept
of the correlation function for two characteristics of the system f, g at points x, y and for the different times.
At comparatively large time after a relaxation the events become independent
and the state of the system becomes equilibrium. The difference between the
left and the right parts characterizes the deviation of equilibrium.
Nonequilibrium
statistical mechanics base
From the point of view of the nonequilibrium
statistical mechanics the general balance equations are not entirely localized
under essentially nonequilibrium conditions. The most profound result obtained
in scope of the nonequilibrium statistical mechanics by the method of the
nonequilibrium statistical operator (Zubarev D.N.) consists in the general
governing relationships between the conjugate thermodynamic fluxes J and forces G (gradients of macroscopic variables) valid through the relaxation
regimes. In scope of the general relationships the dependence of the stress
tensor components J on the
strain-rate tensor components 
has an integral form
. (1)
The first unit in the relationship (1) presents
the reversible part of the flux corresponding to the quasi-equilibrium
distribution function in particle velocities. The second unit presents the
irreversible part of the flux due to the collective interaction effects.
The main meaning of the relationship
(1) is connected with the fact that nonequilibrium statistical ensemble at the
microscale gives rise to another nonequilibrium statistical distribution 
at the intermediate scale between micro- and macro- levels.
This rapidly changing during high-speed deformation distribution plays a role
of a weight function that determines the contributions of mesoscale phenomena into the
macroscopic medium behavior. This
space-time distribution function of correlation among the medium elements up to
the normalization conditions presents the same correlation function mentioned
in the previous section. It means that a state of the system in a point r at an instance t is generally determined only when all the history of the system
evolution all over the space region full of a medium is given. Further, it will
be shown that the integration of the strain-rate over the coordinates
introduces the dependence of the stress tensor on the deformation tensor. Then
it will also be demonstated that Eqn (1) extends the well-known Maxwell model
of the viscous-elastic-plastic medium for the high-strain-rate conditions. From
the general properties of the correlation functions one knows their behavior in
the limiting cases.
At the initial instance t=0 the system had been subjected to an
external loading during a time interval tR , the typical stress relaxation time for the
given medium and the loading conditions is tr.
The relaxation proceeds by stages. The initial stage of frozen relaxation
corresponds to the case where the correlation is constant that is defined by
its initial value 
. An example of the system with the fixed correlation is an
absolutely rigid solid at the zero absolute temperature. In real situation the
correlation is not entirely fixed there exist heat fluctuations at
microscale. In this case the dissipation
can be neglected and the transport mechanism should consider reversible. At
this stage the system remembers its prehistory 
where 
has a meaning of the
initial mean strain that the system remembers during the initial stage. The
typical duration of the initial stage determines the form memory properties of
the medium. At the same time while the stress is proportional to the strain the
first stage corresponds to an elastic process and when the stress would exceed
an elastic limit the elastic stage stops and the stress begins depend on the
strain-rate. It corresponds to the conventional concept of the elastic-plastic
medium. Then the initial relaxation stage approximately corresponds to the
condition 
under which the
classical theory of elasticity is valid.
The final near-equilibrium
hydrodynamic stage 
corresponds to the
case where the memory and nonlocal effects can be entirely neglected. In
hydrodynamics usually the other representation in term of the velocity
gradients is used instead of the strain-rate tensor 
where u is the momentum transport velocity.
Then the transport relaxation
kernels determine the transport coefficients in scope of the linear
thermodynamics of irreversible transport processes.
J(r,t) » k0(0)(r,t)
(r,t) º JN-S(r,t), k0(0)(r,t) º 
r¢Â(r,r¢, t,t¢). (2)
At this stage the system forgets its prehistory 
. The transport coefficients k0(0) near the local equilibrium present the medium
macroscopic property of viscosity conditioned by the medium structure at the
microscale. For the momentum transport this limit results the well-known
Navier-Stokes equations that present a theoretical base for the classical
hydrodynamics. As far as a flow velocity grows the viscosity coefficient first becomes
not constant and begin depend on the temperature and pressure making the
problem nonlinear, then it becomes insufficient to describe flows of structured
medium. Navier-Stokes equations are the second order partial differential
equations of the parabolic type describing the diffusive transport mechanism.
So, the limit of the neglected correlations corresponds to a fluid-like medium
behavior.
In general case at the intermediate
stage the space-time correlations are not enough small to be neglected and
cannot consider being constant. The macroscopic medium reaction on an external
loading should be entirely determined by the time evolution of the finite-size
correlations. In this case the combined medium reaction on the external loading
should be expected. The balance equations are integro-differential and describe
the both mechamisms of reversible and dissipative transport. So, Eqn (1)
resulted from the nonequilibrium statistical mechanics confirms that without
taking into account the space nonlocal and memory effects in between the
limiting situations it is impossible to describe the nonequilibrium processes
in a correct way and the gradient-type theories allow only an asymptotic
description near the limiting situations.
The principle difficulty consists in
that one doesn’t know the correlation function behavior except it is a
nonlinear functional of the macroscopic gradients G . If the relaxation kernel
would be determined by the gradients the balance equations entirely determine
the nonequilibrium mass, momentum and energy transport. However, in general
case it is impossible to derive from the initial nonequilibrium ensemble even
approximate closed form of the functional. So, the only way to deal with the
relationships (1) is a modeling of the correlation function on the base of the
general invariance principles and the known asymptotic behavior. However, a
substitution of simple space- time dependencies into the kernels doesn't allow
an adequate description of nonequilibrium processes and a satisfaction to the
natural boundary conditions for the continuum. Many decades this circumstance
was an obstacle for the nonlocal models to apply to nonequilibrium transport in
real media. The relaxation models necessarily should include a functional
dependence on the gradients G connected
to an external loading that determines the medium reaction on the
nonequilibrium conditions. Without this close-loop between the external loading
propagation and the medium reaction on it any description of nonequilibrium
processes would be not only incomplete but incorrect.
Introduction of the
mesoscale as a result of nonequilibrium transport
The existence of the finite-size
space-time correlations means that for a finite time interval there is a
dependence of the point‘s coordinate in a medium on time. Then one can
introduce a trajectory in the coordinate space that determines the momentum
propagation in a medium 
. The equation doesn’t govern a real motion of the
finite-size medium element. The time derivative 
determines the finite velocity of the perturbation
propagation in a medium. In equilibrium this velocity 
tends to the sound
velocity. In this case there is no mass transport and no dissipation that follows
it.
So, the integration
over space and time in Eqn (1) resulted from the nonequilibrium statistical
mechanics can be reduced to an integral either over space or time and one get
(3)
The integration in Eqn (3) is going
over the volume 
embraced by the
perturbation that has already come from a boundary 
through out which the pulse and energy fluxes input.
Resolution of the function G by
Tailor near the point 
and its substitution
into Eqn (3) result an infinite order differential operator
(4)
The first moments of the
nonequilibrium statistical distribution of the space nonlocal correlation have
definite physical sense related to the medium internal structure. The 0-order
moment k0(r,t)=k0(0)(1+a)
defines effective values of the transport coefficients for the medium
with internal structure. The 1-order moment k1(r,t)
defines a mathematical expectation of the vector 
that is not equal to
zero by definition if the statistical distribution differs from the normal one.
There appear new direction and new length defined by the vector 
generated by
nonequilibrium transport in a real medium. It means that any nonequilibrium
distribution gives rise to the medium polarization along the direction of the
vector under an external
loading through the medium boundaries. This vector determines a shoulder of a force acting on a finite size medium element in the
inhomogeneous velocity and stress fields. It means that such an element should
rotate as a whole and can be considered as an element of the internal medium
structure. The resulted structured medium is characterized by internal
rotations and asymmetrical stress tensor.
In general case the resulted structure size can be arbitrary. The
experiments (Mescheryakov) on the shock wave propagation in solid materials
demonstrate the structure formation at the mesoscale that is intermediate
between micro- and macro- scale levels. At present the mesoscale phenomena
consider being related to the grains and subgrains defect structures in
crystalline materials (Panin) However, the experiments(Sudenkov)
show that new structures formed under dynamic loading can go through the grain boundaries
independently on interfaces. But under another loading
conditions, as other experiments show a grain begins to rotate as a
whole. It means that under definite conditions of external loading a medium
demonstrates its internal structure effects only at definite scales by a
resonance way. And then one can call real mesoscale phenomena only those that
are responding to an external loading in a resonance way.
The second
moment k2(r,t)=
determines the dispersion of the space correlation
distribution 
when 
and introduces the
share of the nonlocal effects as a degree of the deviation from the local
equilibrium state of the system. Really, in the case where 
the nonlocal effects
can be neglected, the distribution function tends to the d- function. This case corresponds to the fluid
reaction to perturbations near equilibrium. In the opposite case where 
the distribution
doesn't decay with the distance from the point r remaining constant for
any finite product
. In this case the
medium in a whole is embraced by correlation propagating from the loading
region. that corresponds to the elastic reaction of a
solid. In both limits the medium doesn't demonstrate the internal structure effects, its behavior corresponds to the structureless
response in scope of the continuum mechanics validity.
In the intermediate case 
under a pulse loading
the medium reaction is like one of a multi-phase dispersed mixture of solid
phase grains with rigid correlation and viscous liquid with weak correlation.
The share of solid phase depends on the medium properties in the initial state
and on the regime of the external loading. Besides, it can rapidly evolve with
time.
Self-consistent
nonlocal theory of nonequilibrium transport
On the base of the fundamental
result of nonequilibrium statistical mechanics a new self-consistent nonlocal
theory of nonequilibrium transport has been developed [ ]. In scope of the theory the space nonlocal
and memory effects prove to be connected due to the finite space-time
correlations and the finite rate of the perturbation propagation in real
medium. Then one needs either nonlocal models or memory ones to construct in
order to get a closed description of nonequilibrium transport. Both type models
differ in general but lead to the same limits. In order to describe the
internal structure evolution the space nonlocal model form is chosen.
Then, the
nonlocal model for the correlation function depending only on the first moments that have a
meaning of the mesoscale structure characteristics is introduced.. The first moments approximation for the
correlation function results a new governing relationship between the stress and the
velocity gradient tensors with the medium structure parameters included.
. (5)
Here the function w determine a type of the space correlation
decaying with a distance from the point 
. The higher moments of the statistical distribution in Eqn
(5), characterizing fluctuations of the medium structure parameters are not
disappeared in the model and can be expressed through out the first moments.
Following an analogy with the well-known local-equilibrium statistical
distribution in velocities the nonlocal distribution becomes normal near the
hydrodynamic limit of the weak nonlocality near the local equilibrium while for
growing values of the nonlocal parameter the function w(|Z|) can take assymmetrical form. However, at the
rapidly changing model parameters the main problem consists in the behavior of
the medium structure parameters and not in the form of the function
w.
There arises a problem for the nonlocal model
(5) of the model parameters determination taking into account that the
relaxation kernel should be a nonlinear functional of the gradients G and the
external loading conditions. The only real possibility for the model is a
definition of the structure parameters as being functionally depending on G. Then the structure parameters can be
determined by using boundary and integral conditions imposed on the system
including the external loading. At arbitrary model parameters it is impossible
to satisfy these conditions that are natural for a continuum description. So,
the boundary and integral conditions can be satisfied only on account of the
parameters and therefore determine them.
(6)
According to Eqn (6) the relationships for the
space nonlocal models determine the structure parameters 
as time-depending functions while for the memory models, to
the opposite, should depend on coordinates.
The nonlinear relationships (6)
complete the model and due to their nonlinearity determine the time evolution
of a spectrum of internal structure scales. In general nonequilibrium case the
scale spectrum is discrete and in limiting cases it becomes continuous defining
the validity region of continuous mechanics. The bifurcation points for the
branching equations (6) point out the structure transitions in the system and
determine the threshold values of the loading parameters. It means that the
condition (6) allow the self-organization effects included into the nonlocal
model.
It is very important to note that
the governing equation (5) on the conditions (6) complete the balance equations
in a self-consistent way. It means that macroscopic fields in a medium depend
on the medium internal structure and the structure evolution is determined by
the macroscopic fields and external loading conditions. There arises a
close-loop in a nonequilibrium system that introduces a self-regulation
included into the nonlocal model.
So, the self-organization and
self-regulation appear to be the necessary components in order to make the
nonlocal model complete and to describe nonequilibrium transport in a correct
way.
The shock induced pulse propagation in a relaxing medium
In scope of the new self-consistent nonlocal
approach for the plane pulse of the moderate intensity propagating along the
x-axis, the
governing relationship (5) takes a form
.
(7)
Here u=dx/dt presents a velocity of the
momentum propagation For pulses of the
moderate intensities the velocity u can be devided into two parts: 
where
is a velocity of the wave propagation that is considered
being constant only near the limiting case of equilibrium, and 
is the mass velocity
considered comparatively small 
. It must be noticed that here 
is only the irreversible part of the mass flux
while the reversible one connected to the particle displacements inside the
wave (elastic deformation) disappears after the front. It is here the main problem arises: how to
devide a deformation into elastic and plastic parts or the velocity into the
wave propagation and the mass velocity in a correct way? In the general case it
is impossible because of a shortage of the needed for this purpous balance
equations. So, the concepts of the mass and the wave propagation velocities
have their conventional meaning only in the case where . It is also means that the concept of the
viscous-elastic-plastic medium cannot be used for strong shock description.
Considering the parameter 
small and 
the mass and momentum
balance equations in the linear approximation are written as follows
=0 ;
(8)
. (9)
Here 
, the ratio 
i s considered being of the same order as the parameter .
At the initial stage of the relaxation when considering 
and 
in the limiting case 
, 
, Eqn (7) results
(10)
Here
Cl2=C2+(4/3)Gr0-1 is
the longitudinal sound velocity and G
denotes the elastic shear modulus. The first unit of this relatioship presents
an equation of solid state 
that at the initial instant is in equilibrium. The second one
describes an elastic medium reaction to an external loading. At this stage the
stress in (10) depends only on the strain, not on the strain-rate. The direct
proportionality between stress and strain means that the medium behavior at the
initial stage corresponds to an elastic solid independently on the medium internal
structure and its nature. Such a behavior can be observed, for example, for
liquids under strong shock. However, the duration of this stage can essentially
vary for different kind media.
The mass and momrntum balance
equations (8)-(9) with the governing relationship (7) can be reduced to a one
wave type equation
0 . (11)
So, the initial stage of relaxation described
by Eqn (11) corresponds to a reversible wave-type mechanizm of the momentum
transport. It must be noticed that during the initial stage there is no
dissipation of the mechanical energy into a heat. It had been experimentally
found out that for high-strain-rate processes the formation of the dissipation
mechanizms require a finite time exeeding microscale but much less macroscale
and depending on material properties.[Rosakis]
At the final hydrodynamic stage when the another
limiting case can be considered 
, 
Eqn (7) results the newtonian model of the medium that
presents a theoretical base for the well known hydrodynamic Navier-Stokes
equations
(12)
At this final stage the stress
consists of two parts: reversible isotropic P
, hydrostatic pressure and irreversible bulk and shear viscous stress. The
pressure 
is connected to the reversible wave-type momentum transport.
For the loading lower the elastic limit Eqn(12)
includes only the reversible component. The balance equations (8)-(9) in this
case result the wave equation with the bulk sound velocity propagation C instead of the longtitudinal one Cl
(13)
The irreversiuble part depends only
on the strain-rate but not on the strain itself. It means that for strong and long-time
loading all media should demonstrate the fluid-like behaviour. Certainly, for
solids this stage can
be reached at very long times. Even brittle materials can flow at a very small
velocitiers like glaciers, for example.
At this stage for tyhe loading higher the elastic limit Eqn(9) with the relationship (12) independently on the mass
balance results a diffusion-type equation
(14)
It means that at the final stage the
momentum transport is going only via the diffusion transport mechanizm.
At the intermediate stage of relaxation when 
and the nonlocal and
collective effects cannot be neglected the momentum transport combines both
wave and diffusion transport mechanizms and is described by the equation
, (15)
where the right hand part is defined by
the nonlocal model.(7 ). At the initial stage it is reduced to the wave Eqn
(11), and at the final hydrodynamic stage it can be shown being diffusive type.
Thus, Equ (15) change its type from the hyperbolic one through the
integro-differential equation and up to the parabolic type at the final
hydrodynamic stage depending on the model parameters evolving with time. As the
parameters are closely connected to the medium internal structure it means that
the structure evolution defines the relaxation or it is better tot say the
medium reaction to the external loading. In the limiting cases the structure effects desappear and the medium behavior corresponds to the
structureless medium described by the conventional continuous mechanics. So, in
order to describe the medium reaction to the high-strain-rate it is necessary
to involve the internal structure evolution.
Maxwell relaxation theory
The well-known Maxwell model of a medium can determine the stress
component J:
,
(16)
where v is the mass velocity in the x
direction, , and 
is the relaxation
time. The medium is assumed to have a well-defined critical shear stress, J*. Here, instead of the strain, the
velocity v is chosen in our
description as a variable that can be measured in experiments on shock
compression of materials. Eqn (16) as a linear 1st order usual
differential equation, when constrained by the initial condition J(0) = 0,
integrates to the expression
. (17)
Eqn. (17) results that the elastic, plastic and viscous medium
properties can be specified only relative to the loading time or to the time typical
for the process compared to the relaxation time. This is the very important
sequence of the Maxwell theory.
For large times compared to the relaxation
time ( t>>), in the limiting case of complete relaxation,
Eqn. (17) yields the viscous-plastic medium model with the Newtonian
viscous term:
.
In the absence of dissipation, the last term can be neglected and, over
the threshold J*, an ideal fluid flow
takes place. This is the ideal plasticity model.
In the case of frozen
relaxation for short time intervals ( t<< ), the elastic solid model is obtained:
.
,
For the time intervals compared with the
relaxation time ( t ~ ),
the first term in Eqn. (14) is growing in time from zero and reaches the value J* only for large time. It means that
agreed-upon classification for elastic, plastic, viscous or viscous-elastic,
elastic-plastic media can be used only on moderate loading conditions and at
rather larg time when the state of the system is near-equilibrium one. If the
loading conditions vary rapidly this classification becomes invalid. In scope
of the theory even the main medium properties such as solidity or fluidity can be
correctly determined only with respect to the loading regime applied to the
medium. Besides, the model being additively constructed doesn’t assume any
collective interaction mechanisms in the intermediate region between the
limitting solid and fluid states ( t~ ). At present it has been found out that namely in this region the
mesoscopic structures occur.
In the case where the
relaxation time 
is constant, Eqn. (5.2) takes the form
(18)
The integral kernel in this relationship is simple exponent with the
relaxation time as a parameter. It becomes clear that all the restrictions
concerning the validity region of the Maxwell theory are connected to the
integral kernel form. The more the integral kernel differs from the simple
exponent, the wider the validity range of the model and the more high-rate
processes it can describe.
Stochastic fluctuations at mesoscale
and mass velocity dispersion
In paper [Rosakis] it had been experimentally demonstrated that the
work of the form changing was not immediately going into a heat and there
should be another process except dissipation that could contribute into the
internal energy in order to satisfy the energy balance equation during the
high-rate straining. Mescheryakov et al. [–] had experimentally found that the
stress relaxation in the velocity range 500–1000 m/s
was followed by formation of mesoscopic structures. During this process a part
of the macroscopic kinetic energy within the wave front is going to large-scale
velocity fluctuations of the medium elements at mesoscale. The quantity characterizing
the velocity fluctuations that can be experimentally measured and immediately
included in the balance equations is the mass-velocity dispersion, D2(x, t)
, originating because of the macro–meso energy exchange. Then D(x, t)
has a meaning of an averaged velocity at mesoscale and D2 has a
meaning of a nonequilibrium temperature. Near equilibrium when the mesofluctuations
become the heat ones the temperature take its conventional meaning. As
experiments show the velocity dispersion is generated before the temperature of
the deformed medium begins growing. This means that the energy exchange between
the macroscopic and mesoscopic scale levels during high-rate straining begins
before the dissipation into a heat (to the microscopic or atom/dislocation
scale levels). First, the mass-velocity dispersion grows then they can decay or
stay constant. During high-rate straining
the fluctuations can change their scale. They aren’t heat fluctuations and
therefore not submitted to the second law of thermodynamics.
Supposing the fluctuations at
mesoscale described by the velocity dispersion being reversible and isitropic the state
of a system at mesoscale is considered as equilibrium. It means chaos at
mesoscale or zero-mean centered noise with respect to macroscale. As soon as
the sychronization of fluctuations begins and the averaged over mesoscale
ensemble velocity is not zero the collective interaction ceases being isotropic
and contributes to macroscale. Then the reversible mesofluctuations are
originating still at the initial stage and prologing their evolution at the
intermediate one also as reversible.
Introducing the stochastic isotropic
component of the mass velocity into the relationship (7) on the condition corresponding to the
initial stage it is possible to get
(19)
So, we get a new nonequilibrium
equation of state for the spheric part of the stress that valid under the
condition 
.(20)
At the final stage of relaxation the
dissipation begins, mesofluctuations become heatfluctuations of atoms at
microscale and the Eqn (19) takes its conventional form derived in the kinetic
theory and equilibrium thermodynamics.
Energy balance and the nonequlibrium
temperature
First it is necessary to define the
internal energy in the case of moderate loading connected to the fluctuations
at mesoscale.
(21)
where cm
is an energy capacity of mesofluctuations and 
is the diffusive
velocity loss at mesoscale due to the irreversible strucrure formation . So,
the first unit of the internal energy is the mean kinetic energy at mesoscale
and the second one is the potential energy provided in irreversible structures.
The energy balance determines
multi-stage and multi-scale energy exchange where the mesoscale plays a role of
an energy buffer between the macro- and micro- levels.
At the initial stage t<< tr
mesoscale is still frozen cm®¥ , the
velocity dispersion does’t grow (D=D(t=0),
the energy balance at microscale corresponds to reversible transport processes.
At the last hydrodynamic stage t>>tr mesoscale is already relaxed cm®0 , the energy of mesofluctuations has
converted into a heat. At the intermediate stage the energy exchange can be
both reversible at finite times and irreversible depending on whether or not
new irreversible structures had formed inside the wave front ¶Σ¤ ¶t=0.
Eqn (22) expresses the physical nature of the collective effects at
mesoscale that is related to the cascade processes followed by both the scale
splitting and self-organization, reversible and irreversible transport and
transitions between different scale levels.
Nonlocal automodel solution
In the reference connected to the
elastic precursor 
Eqn (15) in scope of
the linear approximation with respect to the small parameter 
results
(23)
Eqn (23)
is transformed into the time representation, that more convenient for
experimental measurements of the velocity profile in real time. The Eqn (23)
should be completed by the internal energy balance equation determining the
velocity dispersion. In the case where the irreversible structure formation is
absent and the meso-macroscale exchange is reversible the energy balance can be
written in the integral form
(24)
The energy balance (24) in the local form
requires an assumption that the macro-meso-energy balance is reversible at any
time inside the wave front while the Eqn (24) asumes the reversibility of the
process only after a finite time equals to the rise-time
past.
(25)
It means that
the kinetic energy lost due to the form changing of the wavefront goes
to the mesoscale as the reversible mesofluctuations inside the front and comes
back to macroscale at the pulse plateau. Then the nonsteady velocity profile is
governed by the following integro-differential equation
(26)
Eqn (26) due to the smoothing integral
operator describes a spreading out of
the front during its propagation because of the mesofluctuations like the wave
packet However, it must be noticed that
Eqn (26) is valid only on the condition of the equilibrium macro-meso-energy
exchange inside the wave front. In the elastic regime Eqn (26) converts into
the identity. In the hydrodynamic regime Eqn (26) results an exponential form
of the velocity profile
. (27)
So, two conquring processes show to be involved
in Eqn (26): the smoothing of the velocity profile when the initial pulse is
relaxing and partially dispersing due to the mesaofluctuations at the fixed
structure parameters and the front
steaping due to the structure evolution to hydrodynamic regime near
equilibrium. It means that a temporal balance between the two processes is
possible and a steady front can exist.
In the nonlinear approximation where
the convection is taken into account and the dispersion isn’t suppressed by the
dissipation the obtained equation can result the solutions describing the
propagation and relaxation of several solitons.[ ] Near the hydrodynamic limit it would
transform into the well-known Corteveg-de-Vriese equation [ ]. Near the elastic limit the nonlinear equation
describes the dispersion of the initial pulse [ ]. For the intermediate
mesoscale region mathematical methods based on the iteration procedure
developed specially for such class nonlinear operator sets can be used to get a
solution with any needed precision [Vavilov].
Structure evolution
In order to complete Eqn (26) it is necessary
to add the integral conditions (6) and to define the structure evolution
following the pulse propagation. However, the principle problem arises
connected to the information about the dependences between the mesoscale
evolution and the integral macrocharacteristics behavior. For this reason
another methods of cybernetic physics can be applied [Fradkov]. Last decade
such approaches based on variational principles related to the entropy
production criteria begin developing [ } The integral entropy production inside
the wavefront can be written as follows
. 
(28)
The point above 
denotes a derivative
with repect to the variable 
along which the
integral entropy production evolutes. The stochastic velocity component
connected to the reversible mesofluctuations don’t contribute into the entropy
production. In the elastic regime the entropy production looses its meaning and
becomes an elastic energy 
, and in hydrodynamic limit it results
As an integral value the entropy production
(28) changes much slowly than the front inside. This assumption corresponds to
the scale splitting 
that had already used
in Eqn (26). Taking the integral entropy production as a goal functional and
using the speed-gradient method [Fradkov] it is possible to define the
structure evolution by the following set of nonlinear differential equations
for the structure parameters 
.
(29)
The set (29) describes the structure evolution
in accordance with the leading of the entropy production to its minimal value.
The set (29) depending on the velocity profile itself introduces a close-loop
into the system and makes it complete and self-consistent. Due to the condition the feedback becomes
long-range that provides the necessary condition for synergetic
self-organization in the system. The structure formation at mesoscale in the
form both shear and rotations had been found out in experiments on the shock
loading of differentmetals.[Mescheryakov]
The bifurcation instances for the nonlinear set
(29) determine the structure transitions occuring inside the wavefront and
typical living time for the structure of fixed size depending on its initial
size and the loading regime. [Nikulin]
So, the self-regulation and self-organization
should necessarily be involved into the system in order to complete the balance
equations under nonequilibrium conditions that take place during the shock
loading of real media. In cybernetics it had been demostrated that the system
with the feedback is much more stable with respect to external perturbations
than systems with fixed programme.[Granichin]
It should be underlined that the concept of the medium structure itself
exists only under the condition that structures live longer than risetime.
If to define the rate of the entropy production
(28) and to take into account the structure evolution (29) one can easily get
(30)
The relationship (30) corresponds to the
general principle of evolution introduced by Glansdorf and Prigozgin. The princiole (30) cxharacterizes a general
tendency of the system to equilibrium at
the relaxation stage that can proceed not only due to the dissipation but also
on account of a wave dispersion on the reversible mesofluctuations. The
irreversible structure formation inside the wavefront or during the loading
itself can violate the principlle (30)
and lead to changing of the sign in (30) corresponding to the self-organization
in the system.
References (to be revised! )
1. Vladimirov V.I., Romanov A.E.
Disclinations in crystals. Leningrad. Nauka, 1986, 224p. (in Russian).
2. Co-operative deformation processes
and strain localization. Kiev, Naukova Dumka, 1989, 320p. (in Russian)
3. Rybin V.V. Large plastic deformation
and failure of metals. Moscow, Metallurgia, 1986, 224p. (in Russian)
4. Nicolis G., Prigogin I.
Self-organization in non-equilibrium systems. Moscow, Mir, 1979, 512p.
5. Highly excited states in solids. Ed. V.E. Panin. Izvestiya vuzov. Fizika, N1, 1987,
126p. (in Russian)
6. Hutchinson J.V., Evans A.G.
Mechanics of materials: top-down approaches to fracture. Acta Mater., 2000, 48,
pp. 125-135.
7. Needleman A. Computational mechanics
at the mesoscale. Acta Mater., 2000, 48, pp. 105-124.
8. Aifantis E.C. Gradient aspects of
crystal plasticity at micro and macro scales. Key Eng. Mat., 177-180, 2000, pp.
805-810.
9. De Borst R. Non-local approaches in
mechanics of materials.
10. Richardson.J.M. The hydrodynamical
equations of a one component system derived from nonequilibrium statistical
mechanics. J. Math. Anal. Appl., 1960, N1, pp. 12-60.
11. Piccirelli R. Theory of the dynamics
of simple fluids for large spatial gradients and long memory. Phys. Rev., 1968,
V. 175, N1, pp. 77-98.
12. Zubarev D.N. Nonequilibrium
Statistical Thermodynamics. Moscow, Nauka, 1971. (jn Russian).
13. Zubarev D.N., Tischenko S.V.
Nonlocal hydrodynamics with memory. Physica, 1972, V.59, N2, pp. 285-304.
14. Zubarev D. N. Modern methods of
statistical theory of irreversible processes. Itogi nauki i tekhniki. Seria “Sovremennye problemy matematiki.”,Moscow:
VINITI, 1980, V.15, pp. 128-227. (in Russian)
15. Rudyak V.Ya. Statistical theory of
dissipative processes in gases and liquids. Novosibirsk, Nauka, 1987, 272p. (in
Russian)
16. Filippov B.V., Khantuleva T.A.
Boundary problems of nonlocal hydrodynamics. Leningrad, Leningrad State Univ.,
1984, 88p. (in Russian)
17. Fursenko A.A., Khantuleva T.A.,
Makarov Yu.N., Ramm M.S. Mathematical
simulation of transport processes of species in gas phase in modern
sputter-deposition technologies of thin films. 1.Formulation of the problem. -
A.F.Ioffe Phys.Tech.Inst., Preprint 1595, St.Petersburg, 1992.
18. Vavilov S.A. Geometric methods of
studying the solvability of a class of operator equations. Russian Acad. Sci.
Dokl. Math., 1992, V.45, N2, pp. 276-280.
19. Vavilov S.A. A method of studying
the existence of nontrivial solutions to some classes of operator equations
with an application to resonance problems in mechanics. Nonlinear Analysis.,
1995, V.24, N5, pp. 747-764.
20. Khantuleva T.A., Mescheryakov Yu.I.
Nonlocal effects in the high-strain-rate processes in solids. - in Book: High
Pressure Science and Technology., ed. M. Trzeciakovski, 1995. (Proc.Joint15th
AIRAPT and 33th EHPRG Intern. Conf., Warsaw).
21. Khantuleva T.A., Mescheryakov Yu.I.
Nonlocal theory of the high-strain-rate processes in a structured media. - Intern. J. of Solids and Structures, 1999,
V.36, pp. 3105-3129.
22. Khantuleva T.A., Mescheryakov Yu.I.
Formation of mesoscopic structures in problems on penetration and spalling. -
Chem. Phys. Reports, 2000, V. 18 (10-11).
23. Khantuleva T.A. Non-local theory of
high-rate processes in structured media. –CP505, Shock Compression of Condensed
Matter- 1999 ed. M.D.Furnish, L.D.Chhabildas, and R.S.Hixon, 2000, APS
1-56396-923-8/00, pp. 371-374.
24. Khantuleva T.A. Non-local theory of
high-rate straining followed by structure formations. –J. Phys. 4 France, 2000,
V. 10. EDP Sciences, Les
Ulis, pp. 485-490.
25.
Khantuleva T.A. Microstructure formation in the framework of the
non-local theory of interfaces. – J. Materials Phys. and Mech. 2000, N 2,
pp.51-62.
26. Khantuleva T. A. Non-local theory of
macro-meso-level energy exchange in the shock compressed matter. CP620, Intern.
Conf. “Shock Compression of Condensed Matter”, Atlanta, USA, June 2001, ed. By
M.D. Furnish, N.N. Thadhani, Y. Horie, 2002, Amer. Inst. Phys.
0-7354-0068-7/02, pp 263-266.
27. T.A.Khantuleva. The shock wave as a
nonequilibrium transport process, in monograph “High-pressure compression of
solids VI: old paradigms and new challenges”(Y.Horie, L.Daison, N.N.Thadhani,
Eds.), Springer, Berlin, 2003, pp
215-254
28. Yu.I. Mescheryakov and A.K. Divakov,
“Multi-scale kinetics of microstructure and strain-rate dependnence of materials”,
DYMAT Journal 1, p. 271, (1994).
29. Yu.I. Mescheryakov, A.K. Divakov,
and N.I. Zhigacheva, “Role of mesostructure effects in dynamic plasticity and
strength in ductile steels”, Mater. Phys.
Mech. 3, pp. 63–100, (2001)
30.
Yu.I. Mescheryakov, “Mesoscopic effects and particle velocity
distribution in shock compressed solids”, in: Shock Compression in Condensed Matter—1999,
(ed. M.P. Furnish, L.C. Chhabildas, and R.S. Hixson), American Institute of
Physics, New York, pp 1065–1070,
(1999).