Influence of Inhomogeneity and Initial Stress on the Transient Magneto-Thermo-ViscoElastic Stress Waves in an Anisotropic Solid

The object of the present paper is to study the transient magneto-thermo-visco-elastic stresses in a non-homogeneous anisotropic solid under initial stress. The system of fundamental equations is solved by means of a dual reciprocity boundary element method (DRBEM). In the case of plane deformation, a numerical scheme for the implementation of the method is presented and the numerical computations are presented graphically to show the effects of initial stress and inhomogeneity on the displacement components and thermal stress components.


Introduction
An increasing attention is being devoted to the interaction between magnetic field and strain field in an initially stressed anisotropic viscoelastic solid due to its many applications in modern aeronautics, astronautics, earthquake engineering, soil dynamics, nuclear reactors and high-energy particle accelerators.In recent years, an important number of engineering and mathematical papers devoted to the numerical solution have studied the overall behavior of such materials.El-Naggar, et al. [1,2] proposed explicit finite difference scheme to obtain thermal stresses in a non-homogeneous media.The boundary element method is well known for its accuracy and efficiency in stress analysis (see, for example, Brebbia and Nardini [3], Wrobel and Brebbia [4], Partridge, et al. [5], Divo and Kassab [6], Gaul, et al. [7], Matsumoto, et al. [8], Fahmy [9][10][11], Davi and Milazzo [12]).
The idea of the present paper is to study the transient magneto-thermo-visco-elastic stresses in a non-homogeneous anisotropic initially stressed solid.The formulation is tested through its application to the problem of a solid placed in a constant primary magnetic field acting in the direction of the z-axis.The governing equations are solved by means of a dual reciprocity boundary element method (DRBEM) and then numerical calculations are made for the temperature, displacement components and thermal stress components.The validity of DRBEM is examined by considering a magneto-thermo-viscoelastic solid occupies a rectangular region and good agreement is obtained with existent results.The results indicate that the effects of initial stress and inhomogeneity are very pronounced.

Formulation of the Problem
Here, we present the basic equations of the theory of magneto-thermoviscoelasticity, which will be used for the solution of the problem described above.With reference to a Cartesian frame denoted by 0xyz, consider an initially stressed anisotropic solid is placed in a constant primary magnetic field H 0 , acting in the direction of the z-axis.Here we address the generalized two-dimensional deformation problem in xy-plane only, therefore all the variables are constant along the z-axis.In the 0xy plane, the solid occupies the region which is bounded by a simple closed curve C.
In non-dimensional form the governing equations of magneto-thermo-viscoelasticity for an anisotropic solid can be written as follows (see Fahmy and El-Shahat [13]) , ,  , , , The initial and boundary conditions for the current problem are assumed to be written as , , , , for , where 1 x and 2 x are the same expressions x and respectively, A superposed dot denotes differentiation with respect to the time and a comma followed by a subscript denotes partial differentiation with respect to the corresponding coordinates.
We focus our attention to the case of inhomogeneity along x-axis, so we characterize the elastic constants pjkl C , magnetic permeability  , initial stress P , ther-mal conductivity coefficients pj k and density  of non-homogeneous material by where pjk , In terms of the definitions ( 12) and ( 13), Equations ( 1) and ( 5) can be written as follows With the heat flux vector p q given by Fourier's law where pj It is usually difficult to find out the analytical solutions of such state equations except for some special cases.
k and are taken to be constant in .c R

Numerical Implementation
The main objective of the numerical implementation is to describe the implementation of the DRBEM formulation for the solution of the Equations ( 14) and (15).A more extensive historical review and applications of dual reciprocity boundary element method may be found in Brebbia, et al. [14], Partridge and Wrobel [15], Nardini and Brebbia [16], Albuquerque, et al. [17].

Temperature Field
Weighting (15) with a test function yields Applying Gauss's theorem, integration by parts, Green's second identity and sifting property of the Dirac distribution, we obtain the representation formula According to the DRBEM, the surface of the solid has to be discretized into boundary elements.In order to make the implementation easy to compute, we use collocation points on the boundary and another i in the interior of so that the total number of interpolation points is According to Cho, et al. [18], a function of the form     x k  in the xyplane.
In the xy -plane a spline is of the form where r x    is the Euclidean distance between the field point x and the load point  .and w is a polynomial of degree .For these are the thin plate splines (TPS) Thus, the particular solution of the temperature can be approximated as where Consequently, the dual reciprocity representation formula can be written as follows In which all domain integrals have been transformed to the boundary.
The field variables and are then approximated by means of shape functions where , and T  q  T Φ are matrices We can also approximate the particular solutions and on the boundary as the unknown field variables by means of nodal values and as follows where and are matrices q T q q  Using ( 22) and ( 23) and applying the point collocation procedure to (21), we have the following system of equations , , Using ( 25) into ( 24) we have where the matrices Τ  and   contain the particular solutions.
The generalized source term in ( 18) is approximated with a series of given source source terms q f and unknown coefficients q  as follows Then, by applying a point collocation procedure to Equation ( 27) we obtain Which can be substituted into (27) producing where In order to solve the system (29), the nodal vectors are subdivided into known and unknown parts denoted by the superscripts and u k The following matrix equation is obtained from (29): The unknown fluxes   u q t  are obtained from the first row of matrix Equation (32) are expressed as follows Copyright © 2011 SciRes.WJM Making use of (33) we can write the second row of matrix Equation (32) as follows At a time step , Equation (34) can be written in the following form We take the finite difference grids with   as the time step, and use the subscripts to denote the nt discrete time.A mesh is defined by being the time step Using finite difference, we can approximate the temperature as follows Hence, we can write where determined, the remaining task is to solve (14) subject to (6), (7) and (8).

Displacement Field
Making use of ( 2)-( 4), ( 12) and ( 13), we can write (14) as follows When the temperature field is known, the displacement field is obtained by solving (38), where the inertia term p u   , the magnetic term pl , the viscosity term , the temperature gradient p and the initial stress term D T P jp D are treated as the body forces.On the basis of the method of weighted residuals, the differential Equation (38) can be transformed to the integral equation in the following form where * p u is a weighting function and is the approximate solution.
Using the symmetric elasticity tensor   Using Equations (41), ( 42) and (43), the Equation (40) can now be rewritten in the form We define the fundamental solution by the relation If we replace the weighting functions in (44) by , then we have From ( 38), ( 45) and ( 46), the representation formula may be written as Using the TPS as in Cho, et al. [18], we can write the particular solution of the displacement as follows , which is known as Euler's constant.
Hence, the traction particular solution q pn t and source function q pn f can be obtained by evaluating , , On the basis of these considerations, the integral domain can be approximated as Making use of (47), ( 53) and (54) we can write the dual reciprocity representation formula as follows The representation formula (55) is only valid if  lies inside the domain .To obtain an expression that contains only boundary variables, the load point R  has to be moved to the boundary.Therefore, the boundary is deformed by a small circular region with radius  around the load point C   as shown in Figure 1.
According to references [11], [19] and [20], the dual reciprocity boundary integral equation can be expressed as where is the Cauchy principal value symbol.q (56) The unknown field variables and the particular solutions are respectively approximated by means of nodal where   k   and Φ are matrices.
On the basis of these approximations, and using the point collocation procedure, the dual reciprocity boundary integral Equation (56) results to the following system of equations

 
By letting We can write (59) as follows The coefficient vector     can be calculated from (48) using the point collocation procedure, which yields Thus, (61) yields the system where By subdividing the nodal vectors into known and unknown parts as follows where the superscripts and denote, respectively, the known and unknown parts k u Hence we can write the system (63) in the following form The unknown fluxes  can be obtained from the first row of (66) as follows With the aid of (67) into the second row of (66) we obtain We can write (68) at time step where Now, we consider an implicit backward finite difference scheme for solving the system of ordinary differential Equation (69), the so-called Houbolt's algorithm is applied to reduce (69) to an algebraic system.To do this, the velocities and accelerations at time step Substituting ( 94) and ( 95) into (93) we have where We apply successive over-relaxation (SOR) as described in Golub and Van Loan [21] to solve the system (96).For every time step the values of u    can be obtained from (70) and (71), respectively.For the case in which , the procedure described in Bathe [22] is used together with the initial conditions to derive 1 and 2 .Lastly, we compute the traction vector using (67).

Numerical Results and Discussion
The present work should be applicable to any magneto-thermo-viscoelastic deformation problem.The application is for purpose of illustration; we do not intend to validate the results in a quantitative way because we have no experimental data at hand; this may be justified because our objective is to introduce a viable numerical technique for studying a model rather than to study any physical behaviors of it (see, for example, Ahmed [23], Kanaun [24] and Monsia [25]).Such a technique was     discussed in Abd-Alla et al. [26][27][28] who solved the special case from this study in the absence of viscosity and inertia.To achieve better efficiency than the technique described in Abd-Alla, et al. [26][27][28], we use thin plate splines into a code, which is proposed in the current study.We extend the study of Abd-Alla, et al. [26][27][28], to include the viscosity interactions and the inertia term.Thus, it is perhaps not surprising that the numerical values obtained here are in very good agreement with those obtained by Abd-Alla et al. [26][27][28].
The example considered by Sladek, et al. [29] may be considered as a special case of the current general problem in the context of the uncoupled thermoelasticity theory.Also, there are alot of practical applications may be deduced as special cases from this general study and may be implemented in commercial finite element method (FEM) software packages FlexPDE 6.
In the special case under consideration, the results of the displacement 1 are plotted in Figure 8 to show the validity of the proposed method.These results obtained with the DRBEM have been compared graphically with those obtained using the Meshless Local Petrov-Galerkin (MLPG) method of Sladek, et al. [29] and also the results obtained from the FlexPDE 6 are shown graphically in the same figures to confirm the validity of the proposed method.It can be seen from this figure that the DRBEM results are in excellent agreement with the results obtained by MLPG and finite element methods, thus confirming the accuracy of the DRBEM.With a view to illustrate the numerical implementation presented earlier, we consider a monoclinic graphite-epoxy as an anisotropic magneto-thermo-visco-elastic solid with the following physical constants: Mechanical temperature coefficient and choosing 60 well spaced out collocation points (N 60)  i  in the interior of the solution domain.The initial and boundary conditions considered in the calculations are In order to evaluate the influence of the inhomogeneity on the displacements and thermal stresses in an anisotropic viscoelastic solid under initial stress, the inhomogeneity parameter is taken to be and for the homogeneous solid, we assume that .
0.5 m  0.0 m  A comparison of the results is presented graphically for the following different cases: the solid line denoted by "A" represents the solution for homogeneous solid in the absence of initial stress , the dashed line denoted by "C" represents the solution for non-homogeneous solid in the absence of initial stress, the dotted line denoted by "B" represents the solution for homogeneous solid in the presence of initial stress and the dashed-dotted line denoted by "D" represents the solution for non-homogeneous solid in the presence of initial stress. .Also, they show the difference among the four cases "A", "C", "B" and "D". Figure 3 shows that the displacement 1 increases and then decreases with the increase of u x .It is noticed that the maximum value happens in the homogeneous solid in the presence of initial stress.
Figure 4 shows that the displacement 2 decreases and then increases for all cases.It is clear from this fig-   diverge in the presence of initial stress.We can see also from this figure that the negative maximum value happens in the non-homogeneous solid in the presence of initial stress.
Figure 5 shows that the thermal stress 11  increases with increasing x for all cases.It is apparent from this figure that the increasing rate is more pronounced in non-homogeneous solid in the presence of initial stress.
Figure 6 shows that the thermal stress 12  increases with the increase of x for all cases.It will be observed from this figure that the increasing rate is more pronounced in the presence of initial stress.
Figure 7 shows that the thermal stress 22  decreases with the increase of x for the non-homogeneous solid, but for homogeneous solid it increases with the increase of x .It is clear from this figure that the increasing rate is more pronounced in the presence of initial stress.
It is clear from all of these figures that the curves of the displacements and and thermal stresses    12  and 22  are closer in the absence of initial stress than in the presence of initial stress.We may also observe from these figures that the initial stress has an important effect on the magneto-thermo-visco-elastic stresses along x -axis through the material constants, thermal constants, magnetic constants and viscosity factor, which are essential parameters to be considered in the design of various devices.Furthermore, while there is no limitation in the solution procedure, all boundary and initial conditions are strongly satisfied.This phenomenon gives clear evidence of a magnetothermostress-focusing effect in a non-homogeneous anisotropic initially stressed viscoelastic solid.From this knowledge of the variation of magneto-thermo-viscoelastic stresses along x -axis in a non-homogeneous anisotropic initially stressed viscoelastic solid placed in a constant primary magnetic field, we can design various viscoelastic solids under magnetothermal load to meet specific engineering requirements and utilize it in meas-urement techniques of thermoviscoelasticity.

Conclusions
The purpose of this paper is to investigate the transient magneto-thermo-visco-elastic stresses in a non-homogeneous anisotropic body.A development of the dual reciprocity boundary element method for solving the system of fundamental equations is presented.In the case of plane deformation, a numerical scheme for the implementation of the method is presented and the numerical computations are carried out for the temperature, displacement components and thermal stress components.The validity of DRBEM is examined by considering a magneto-thermo-visco-elastic solid occupies a rectangular region and excellent agreement is obtained with existent results.The results obtained are presented graphically to show the effects of inhomogeneity and initial stress on the displacement components and thermal stress components.


is the mechanical stress tensor, pj  Maxwell's electromagnetic stress tensor, k is the displacement, T is the temperature, is the initial stress in the solid, are respectively, the constant elastic moduli and stress-temperature coefficients of the anisotropic medium, is the viscoelastic material constant,   is the magnetic permeability, is the perturbed magnetic field, h pj are the thermal conductivity coefficients satisfying the symmetry relation points in the medium,  is the density, is the specific heat capacity of the solid and c  is the dimensionless time, r  is the heat source density and is the Euclidean distance between the field point r x and the load point  .Also, k ,


are constants (the values of pjkl C ,  , P , pj and k  in homogeneous matter) and m is a rational numbers.

u
Integration by parts twice using the divergence theorem of Gauss as in Fahmy[11] yields From (45), one may derive

Figure 1 .
Figure 1.Geometry of the deformed boundary.

(Figure 2
Figure 2 shows the variation of the temperature T along x -axis at various values of the time  .It is noticed that the temperature increases with the increase of x and  .

Figures 3 - 7
Figures 3-7 show the influence of the initial stress and inhomogeneity of the material constants on the displacements 1 and 2 and thermal stresses 11 u u  , 12 

Figure 3 .
Figure 3. Variation of the displacement u 1 with x coordinate.

Figure 4 .
Figure 4. Variation of the displacement u 2 with x coordinate.

Figure 5 .
Figure 5. Variation of the thermal σ 11 with x coordinate.

Figure 6 .
Figure 6.Variation of the thermal σ 12 with x coordinate.

Figure 7 .
Figure 7. Variation of the thermal σ 22 with x coordinate.

Figure 8 .
Figure 8. Variation of the displacement u 1 with time τ for the three methods: DRBEM, MLPG and FEM.