The Mesoscopic Constitutive Equations for Polymeric Fluids and Some Examples of Viscometric Flows *

Constitutive equations for melts and concentrated solutions of linear polymers are derived as consequences of dynamics of a separate macromolecule. The model is investigated for viscometric flows. It was shown that the model gives a good description of non-linear effects of simple shear polymer flows: viscosity anomalies, first and second normal stresses, non-steady shear stresses.


Introduction
Describing viscoelastic behavior of the polymer system, one should distinguish the case of highly concentrated (c > 10%) solutions and melts of long polymers-strongly entangled systems ( 10 e M M  ), where M is the length (in any units) of a macromolecule and e M is the length of a part of macromolecule between the adjacent entanglements [1][2][3], and the case of melts of shorter polymers and half-dilute polymer solutions (c ~ 1% -10%)weakly entangled systems ( 10 e M M  ) [4].The convenient characteristic of a system of entangled linear macromolecules (solutions and melts of polymers) appears to be   on the typical plateau, according to the formula for weakly and strongly entangled systems, correspondingly In this paper we consider the case of weakly entangled systems and formulate a rheological equation of state (RES) that establishes a relationship between the stress tensor, kinetic characteristics, and internal dynamic parameters.At present, a large number of such equations of various complexity is known for polymeric liquids, but, despite various approaches both phenomenological [5][6][7] and microstructural [1,2,[8][9][10][11] ones, the problem how to include specific features of a polymer system into the form of constitutive equations has no complete solution.This is due to both complexity of these systems, which are formed by tangled macromolecules, and mathematical difficulties [11].The information about the microstructure and micro dynamics of the material ought to be incorporated into the present theory of linear and nonlinear relaxation phenomena in polymer systems.An advantage of the micro structural approach is a possibility of studying the relationship between the micro characteristics of a polymer system (concentration and molecular weight of the polymer) and macroscopically observed quantities (viscosity, shear and normal stresses, etc.).In this connection using microstructural concepts it's feasible to formulate a sequence of RES that takes into account new molecular effects in each stage.At [4,9,12] obtained and studied a simple rheological model which can be chosen as an initial approximation in formulating such a sequence of RES.In this work, RES [4] is extended to the case of allowance for the additional corrections caused by intrinsic viscosity and the delayed interaction of a macromolecule with its environment.Realization of this approach involves consequent solution of two problems: formulation of the equations of dynamics for a macromolecule and transition from the formulated equations to RES.The resulting equations can be recommended as the first approximation in constructing a sequence of RES.Comparison of the approach with others, Graessley [1,3] and Doi-Edwards [8] approaches, one can find in [9,12].

Dynamics of a Macromolecule in Flow
The mesoscopic approach to the description of the dynamics of polymer systems is based on the equations of the macromolecule dynamics which cannot be formulated without additional assumptions, namely: 1) A monomolecular approximation of the system.It means that, instead of the entire set of interacting macromolecules in the volume we consider are non-interacting separate macromolecule, moving in the effective medium formed by the solvent and the other macromolecules.
2) The coarse-grained approximation of a macromolecule.It means that, irrespective of the chemical nature of the polymer, the slow motions of the chosen macromolecule can be described as motions of N centers of friction (beads) connected by elastic entropy forces (springs) in a chain.These assumptions lead to the following equations of the macromolecules dynamics [9] in normal modes: where i   and i   are the i-th components of the normal coordinates and velocity, is the mass of a m bead.Here  is the friction coefficient of a bead in a monomer fluid, ij  is the velocity-gradient tensor, which is conveniently expressed below as the sum of symmetric ij  and anti-symmetric ij  parts, and the bracketed expression is the difference between the particle velocity at a given point of space and the velocity of undisturbed flow at this point.The force i   describes the interaction of the polymer chain with the environment via the solvent, while the force i T  is an intrinsic resistance force, i   is a random force, and 2T   is the coefficient of elasticity.
Expression (1) is the basis for the description of the dynamics for different polymer systems [12].The definition of the extra forces i   and i T  allows one to specify the polymer system.Different models of these forces correspond to different physical cases.For dilute polymer solutions, in which polymer macromolecules can be considered non-interacting, the extra forces are considered to be equal to zero [9].In concentrated polymer systems, the macromolecules cannot be considered as not interacting.So, one has to take into account the reaction of the environment and the strengthening of the friction coefficient.The first factor is due to the delayed character of interaction of the macromolecule with its environment, and the second is due to the fact that the chosen bead undergoes resistance not only from the monomer solvent, but also from other macromolecules.Furthermore, one should take into account that, in the flow with nonzero velocity gradients, a macromolecular coil changes its form, and the medium, formed by the coils becomes anisotropic.This mobility anisotropy of beads is called induced and is determined by the shape and orientation of macromolecular coils [4,9,12].According to all these factors the equation for the force of hydrodynamic entrainment follows: Here  is the relaxation time of the environment,  is the dimensionless tensor friction coefficient of a bead, is the strengthening measure of the friction B coefficient  , and is a parameter.The bracketed p expression on the left side of ( 2) is a substantial derivative of the vector quantity i   [9].The presence of this derivative allows one to meet the principle of material objectivity in Equation (2) [6,9].Numerical parameter p entering into the definition of the substantial derivative can take different values.For = 0, the substantial p derivative becomes the Jaumann derivative which has a simpler form while for = 1 and -1, it becomes the p upper and lower convective derivatives, respectively.The specific value of corresponding to one of the p above-mentioned derivatives in (2) is determined below.
If macromolecules form a tangled system besides the force of hydrodynamic entrainment, one should take into account the intrinsic viscous force i T  , the meaning of which is elucidated by Pokrovskii et al. [12].The specific requirement imposed on the force i T  is that this force is vanishing when a macromolecular coil is rotating as a unit.All this allows one to write the equation for this force in similar to (2) manner in the form where: 0 ij  is the dimensionless tensor friction coefficient and is the strengthening measure of the friction ) has relaxation character and depends on the anisotropic properties of the environment.
We assume that the anisotropy of mobility in considered polymer system is characterized by the second-order symmetric tensor .Then, for coefficients Thus, (1-4) is the system of equations of dynamics of a macromolecule.The random force entering into (1) is the Gaussian random process with a zero average.Its correlation tensor satisfies the corresponding fluctuation-dissipation relation [9,10].

Stress Tensor and Rheological Equation of State
Equations ( 1)-( 4) give a microscopic picture of a polymer system flow based on discrete variables.Transition to the continuous case, i.e., to the description of polymer-system flows in terms of continuum mechanics, requires introduction of macroscopic variables-density  
x t m x r Here r  and u  are vectors of position and velocity of the bead with number  , x is the co-ordinate vector of the chosen point in space, and is time.Sum is t taken over all beads in a unit volume, and averaging is performed over the ensemble of all possible realizations of the random force . Differentiating (5) due to time yields an equation of mass conservation, and transformation to generalized coordinates using (1) yields an equation for momentum density.In the latter case, we have an expression for the stress tensor of a polymer system in terms of statistical characteristics of the system (1-4) solutions: where 0 is pressure, n is the number of macromolep cules in a volume unit, T is the temperature in energy units, and 2 3 In the inertia-free case (m = 0), one can formulate (see Appendix) the relaxation equations for the dimensionless correlation moments ik x  and ik u  in the following form where: is a set of the Rouse relaxation times.In Equations ( 8)-( 9) symbols , , , Thermodynamic variables ik x  entering into (6) characterize the inertial properties of a macromolecular coil and, hence, can be used to determine anisotropy tensor ik a in (4).Following [9], we write Therefore it becomes possible to establish the physical meaning of the microanisotropy parameters entering into (4).These parameters take into account dimensions   ,   and shape   ,   of a macromolecular coil in the equations for macromolecule dynamics.
The Equations ( 6), (8), and (9) define a nonlinear, anisotropic, viscoelastic fluid.The behavior of system ( 6), (8), and ( 9) is determined by the six dimensionless parameters ( 2B , and   ) and two dimensional parameters ( B  and ).Parameter nT  characterizing the ratio of the relaxation time of the environment  to the maximum relaxation time B  was estimated in [9], where it was shown that 1   for sufficiently long polymer chains.As to parameter  , here two cases can be distinguished: 1   [13,14-16] and 1   [5], which are discussed below.As in [13], it is convenient to consider simpler forms of equations (8) and ( 9) by using the smallness of the parameters  and  .We consider in more detail the case 1   , which corresponds to the dynamics of polymer solutions at a concentration about of 1%.
Considering only effects of the first order with respect to  and  , we note that Equations ( 6) and ( 8) do not change, and Equation ( 9) takes the form In the zero-th approximation for  and  , variable 0 ik u   and Equations ( 6) and ( 8) take the form 0 1 3 3 The parameters of this system are B  ,  , and  .
Note that when = 1, system (12) is the system of N equations for a dumbbell model where 0 0 nT    and 0  are the initial shear viscosity and the relaxation time, and jj I a  .The system (13) under the assumption of isotropic relaxation (  = 0), is followed by the well-known structural phenomenological Vinogradov-Pokrovskii model [13,[14][15][16][17].
The model ( 13) is simple and gives high accuracy in describing steady nonlinear effects, though it is only the zero-order approximation model, which not permit one to predict all features of polymer flow.In case, one needs more details, one can consider the contributions of parameters  and  which take into account the relaxation character of the environment and the intrinsic viscosity in the equations for macromolecule dynamics.

Linear Effects of the Rheological Model
To obtain an expression for the dynamic shear modulus that corresponds to system (6), (8), and (9), we find a solution to this system in a linear approximation with respect to the velocity gradients.In this case, anisotropy tensor ik is equal to zero, and the terms a ik  can be omitted.Then ( 8) and ( 9) are written as The latter equations can be written as Solving the first equation of ( 14) by the method of successive approximations with first-order approximation due to the velocity gradients, we obtain Substitution of this expression into the second equation in (14) yields In the simple oscillating shear flow    and the last two expressions together with (6) define the Next, it is convenient to distinguish the real and imagi- If the value of the modulus on the plateau is determined by (15), then This series converges only for p = 0. Thus, the Jaumann derivative in the equations of dynamics of a macromolecule ( 2   , which corresponds to the dynamics of melts and strongly concentrated solutions [4], from ( 16) we have   .Using the estimate for obtained in [11], from the last The value of the initial shear viscosity 0  , which can be expressed from ( 16) as To compare the calculation and experimental results, Copyright © 2012 SciRes.WJM we turn to the data of Menezes and Graessley [16,17], where were measured for solutions of polybutadiene with different molecular weights at the same concentration of c = 0.0676 g/cm.The above formulae allow us to find the following estimates of the parameters of the rheological model ( 6), (8), and ( 9

Non Linear Effects in the Simple Shear Flow
The system of constitutive Equations ( 6), (8), and ( 9) should be checked for correspondence to polymer fluid   flows.Diverse flows leads to difficult mathematical problems and, naturally, the check of the RES should be start with the simplest cases.One type of flows that are often realized in viscometers of various design is simple shear.In this case, the velocity-gradient tensor contains only one nonzero component , which varies with time due to a well-known law.In steady flow, the gradient of velocity is constant, and rheological behavior of the polymer system is conveniently characterized by the following viscometric functions: shear viscosity   The dependence  is often given in the form Here  is the shear velocity and   E t is the unit function of Heaviside.In the first case, system (6), (8), and ( 9 Solving ( 8) for case of a simple shear flow with shear velocity 12  with third-order accuracy with respect to Thus, the parameters and   are responsible for the nonlinear properties of system (12).For simple shear,  appears even in the second order with respect to the velocity gradients, and appears only in the third.
 from the condition of the best agreement between the theoretical curves and the experimental data.The value of was not measured in [16,17].To compare RES ( 6), (8), and ( 9) with experiments, we use the data of Menezes and Graessley [16,17], who studied shear flows of polybutadiene solutions with various molecular weights.It is convenient to use their results, because their data on linear viscoelastictity have already been compared with ( 6), (8), and ( 9) in Section 4.
The results of the calculation for the viscometric functions (18) and the corresponding experimental values are given in Figures 8, 9     specimen with are given in Figure 9, which shows that Equations ( 2), (4), and ( 5) describes (the non-monotonic attainment of 5 3. 5 10, which show that (18), ( 8) and (19) confirm the presence of two characteristic times for stress relaxation after intense shear deformation.The anisotropy parameter values are the same as in Figures 8 and 9.

Conclusion
Thus, the proposed microstructural approach to the description of the dynamics of polymer fluids does not contradict the available experimental data on the linear and non linear viscoelasticity of linear polymer solutions and melts.The obtained rheological equation of state is applicable for the description of steady and unsteady effects in both linear and nonlinear regions of strain rates and more complex flow regimes of linear polymer solutions and melts.The model can serve as a basis for the description of nonlinear effects in polymeric systems.

Appendix: Derivation of Relaxation Equations
We introduce relaxation equations for the dimensionless moments ik x  and ik u  .In the inertia-free case (m = 0), Equations ( 1)-( 4) can be written as is a set of the Rouse relaxation times, and is the measure of intrinsic viscosity.
Using (8), we obtain a closed system of equations for the moments which is unknown in (A2), can be found from the fluctuation-dissipation theorem.Besides there is another method.The equilibrium values of the moments were determined previously by Pokrovskii [6]: These values should be obtained from (A2) at zero velocity gradients, and the moments Going over to the dimensionless moments ik x  in (9), we have Since the expression for the stress tensor ( 6) is written in symmetric form, it is more convenient to use the variable   for strongly entangled systems, is inversely proportional to number of entanglements for one macromoleculee M M .Here  is the polymer density.The quantity  can be easily with estimated the value e of real component of dynamic G modulus

1   and 1 
) and (3) corresponds to cases a) and b).The definitions of the relaxation times  , R   , and   are given through parameters  and  , estimates of which are given in[4,9], where it was shown that, for sufficiently long chains, one can always assume 1   .As to intrinsic viscosity parameter  , here two alternative cases  are distinguished.The curves of G   calculated by (18) are given in Figures1-4, from which one can see that the values of by parameter  , and the impact of parameter  (for 1   ) is insignificant.existence of characteristic plateau is determined by relaxation time  .In the case, when  = 0, that corresponds to dilute solutions, irrespective of the type of convective derivative, a plateau on 

Figure 1 .Figure 2 .Figure 3 .
Figure 1.The contribution of parameter  to the dimensionless frequency dependence of real component of dynamic modulus     G  .

Figure 4 .
Figure 4.The contribution of parameter  (internal viscosity) to the dimensionless frequency dependence of imaginary component of dynamic modulus

1 .
) were obtained:  = 0.077; 0.025; 0.011 and 0.005, B  = 0.21; 2.35; 16.27 and 147 sec, and = 840.4;480.2; 321.5 and 206.7 Pa for molecular weights, respectively.In all cases, 025.The comparison of the results is given in Figures5 and 6, from which one can see satisfactory agreement between the theoretical and experimental curves of The values of  as a function of M are given on Figure7showing good agreement with(17).

Figure 7
Figure 7 presents parameter  dependence molecular weight corresponding 1 2 p G

Figure 5 .
Figure 5. Real component of dynamic modulus

Figure 6 .
Figure 6.Imaginary component of dynamic modulus

.
) describes the establishment of stresses from the state of rest, and the corresponding viscometric functions (18) are denoted by In the second case, this system describes stress relaxation after shear deformation, and functions are denoted by of the velocity gradient and time.The following relations are apparently valid

2
Let us consider nonlinear unsteady effects.The results of the calculation of the establishment of stresses for a N At low shear velocities, the second difference of normal stresses is given by the formula 2 6].Thus the model describes a non-zero second normal stress difference.The calculations show that

Figure 8 .
Figure 8. Experimental (solid lines) and theoretical (points) plots of the steady shear viscosity coefficient and the first difference of normal stresses vs. the shear velocity for various values of the molecular weight.

Figure 9 .
Figure 9.The comparison of experimental (solid lines) and theoretical (dashed lines) dependences of an establishment of stationary values of the shear viscosity coefficient and the first difference of normal stresses at various shear velocities.

Figure 10 .
Figure 10.The comparison of experimental (solid lines) and theoretical (dashed lines) dependences of relaxation of the shear viscosity coefficient and the first difference of normal stresses after shear deformation at various shear velocities.
into (A2) in a linear manner.Hence, taking into account the desired moment derivative of the tensor quantity ik x  .Multiplying (A1) by i   and averaging the resulting expression, we write the following equation for ik u  : A5) by multiplying (A1) by k T  and performing averaging.Using (A3), we finally obtain (9)The equation for this variable can be obtained from(9)using the symmetry of tensors ik c  and ik b  and assuming the permutability of tensors ik u  with ik  , ik c  , and ik b  .Assuming the existence functional relationship between ik x  , ik u  and ik  and also by virtue of the fact that ik c  and ik b  are expressed in terms of ikx  , this assumption is not a significant constraint.Then, instead of (9), one obtains