The study of non-Newtonian fluids has generated much interest in recent years in view of their numerous industrial applications, especially in polymer and chemical industries. The examples of such fluids includes various suspensions such as coal-water or coal-oil slurries, molten plastics, polymer solutions, food products, glues, paints, printing inks, soaps, shampoos, toothpastes, clay coating, grease, cosmetic products, custard, blood, etc. Some interesting studies of non-Newtonian fluids are given by Hayat et al. [1-5], Asghar et al. [6], Khan et al. [7,8], Cortell [9,10], Ayub et al. [11-13], Ariel et al. [14], Rajagopal [15-17], Erdogan [18], Siddiqui and Kaloni [19] and Fetecau [20]. Couette flow is an important type of flow in the history of fluid mechanics. Researchers have deep interest in this flow and they study it in many ways. Some important studies about this flow are as follows:

Fang [21] studied Couette flow problem for unsteady incompressible viscous fluid bounded by porous walls. Khaled and Vafai [22] considered Stokes and Couette flows due to an oscillating wall. Asghar et al. [23] discussed unsteady Couette flow in a second grade fluid with variable material properties. Hayat et al. [24] examined the axial Couette flow problem of an electrically conducting fluid in an annulus. Hayat and Kara [25]

studied Couette flow of a third-grade fluid with variable magnetic field. Seth et al. [26] presented Couette flow problem for a porous channel. Bhaskara and Bathaiah [27] have analyzed Couette flow problem for flow through a porous straight channel with MHD and Hall effects. Das et al. [28] considered unsteady Couette flow problem in a rotating system. Ganapathy [29] presented a note on the oscillatory Couette flow in a rotating system. Guria [30, 31] discussed Couette flow problem for rotating and oscillatory flow. Sigh [32] found a periodic solution for oscillatory Couette flow.

The Eyring-Powell model [33] although more mathematically complex, has certain advantages over the Second grade, Maxwell, Power-law and Micropolar fluid models. Eyring-Powell model is derived from the kinetic theory of liquids rather than the empirical relations. It correctly reduces to Newtonian behavior for low and high shear stress. Recently, Eldabe et al. [34] and Zueco and Beg [35] discussed the non-Newtonian fluid flow under the effect of couple stresses between two parallel plates using Eyring-Powell model. Prasad et al. [36] studied momentum and heat transfer of a non-Newtonian Eyring-Powell fluid over a non-isothermal stretching sheet. Patel and Timol [37] presented a numerical treatment of MHD Eyring-Powell fluid flow. Sirohi et al. [38] studied Eyring-Powell fluid flow past a 90˚ wedge. Javed et al. [39] discussed flow of an Eyring-Powell nonNewtonian fluid over a stretching sheet. Noreen and Qasim [40] analyzed peristaltic flow of MHD EyringPowell fluid in a channel.

Keeping this all in view, in the present paper, the authors envisage studying the time-dependent Couette flow of incompressible non-Newtonian Eyring-Powell model with porous walls. The resulting unsteady problem is solved by means of homotopy analysis method (HAM) [41-58], which is very powerful and efficient in finding the analytic solutions for a wide class of nonlinear differential equations. The method gives more realistic series solution that converges very rapidly in physical problems. The convergence region for the series solution is found with the help of. For a given amount of computational effort, one can usually construct a rational approximation that has smaller overall error in given domain than a polynomial approximation [59]. Our goal is to make the maximum error as small as possible. For this purpose, Pade approximants and Taylor polynomials are found. The graphs of the error for Pade approximants and Taylor polynomials are plotted and it is observed that maximum absolute error occurs at the end point. The graphs for the horizontal velocity profile and shear stress at the wall for injection/suction are drawn and discussed in detail. The tables for the initial slope and wall shear stress are also constructed and discussed. More significantly, the series solution clearly demonstrates how various physical parameters play their part in determining properties of the flow.

Consider an unsteady, incomprssible, non-Newtonian, Couette flow problem for the Eyring-Powell model, in which the bottom wall is fixed and subjected to a mass injection velocity and there is mass suction velocity at the top wall, correspond to injection and correspond to suction. The top plate is stationary when, there is only mass transfer in the transverse direction, say direction. At, the top wall is started impulsively to a constant velocity. The Eyring-Powell model is derived from the theory of rate processes, which describes the shear of a non-Newtonian flow. The Eyring-Powell model can be used in some cases to describe the viscous behavior of polymer solutions and viscoelastic suspensions over a wide range of shear rates. The stress tensor in the Eyring-Powell model for non-Newtonian fluids is given by [33]

where is the dynamic viscosity, and are the characteristics of the Eyring-Powell model. Taking the second order approximation of the function as

The governing equation for this problem can be obtained as

where is the kinematic viscosity, is the density of the fluid, bottom wall is located at, top wall is located at and is the velocity at the upper wall. Equations (3) and (4) can be non-dimensionalized by defining

Then Equations (3) and (4) become

where is the Reynolds number, is the fluid parameter and is the local non-Newtonian parameter based on velocity of plate. Using stream function relations with velocity [60] Equations (6) and (7) become

where, is the reduced stream function and prime denotes ordinary derivative w. r. t. When, Equation (8) becomes

where is some arbitrary unknown function of.

To start with the homotopy analysis method it is very much important to choose an initial guess approximation and a linear operator. Therefore, due to the boundary conditions (9) it is reasonable to choose the initial guess approximation

and the linear operator

which satisfies the following property:

where and are arbitrary constants. If is an embedding parameter and is auxiliary non zero parameter then the so-called zero-order deformation equation is

subject to boundary conditions

where

and when and, then

As the embedding parameter increases from 0 to 1, varies (or deforms) from the initial approximation to the solution. Using Taylor’s theorem and Equation (16), one obtains

in which

Clearly, the convergence of the series (17) depends upon. Assume that is selected such that the series (17) is convergent at, then due to equation (16) we have

For the order deformation problem, we differentiate Equations (13) and (14) w.r.t and then setting and finally dividing it by the deformation equation for is given by

where

Following the HAM and trying higher iterations with the unique and proper assignment of the results converge to the exact solution:

using the symbolic computation software such as MATHEMATICA, MATLAB or MAPLE to solve the system of linear equations, (20), with the boundary conditions (21), and successively obtain

The auxiliary parameter gives the convergence region and rate of approximation for the homotopy analysis method for above problem. For this purpose, the is plotted for above problem. It is obvious from Figure 1 that the range for the admissible values for is. The solution series converges in the whole region of and for or.

Pade approximants make up the best approximation of a function in the form of a rational function of a given order. Pade approximation helps us in improving the ac curacy of approximate solution available in the form of a polynomial. Pade approximants are better approximation of a function than its Taylor series, they work even in those cases where Taylor series does not converge. Pade

approximations are also used to enlarge the interval of convergence of approximate series solution [61]. A standard MATHEMATICA routine can be used to find Pade approximant for the function. A Pade approximant for the solution in Equation (24) at, , , , can be written as

Figure 2 depicts the graph of and its Pade approximant. From Figure 2 we observe that the difference between the HAM solution and Pade approximate solution is so small as to be invisible on this scale. The graph of the error

over for the Pade approximant is shown in Figure 3. We note that the maximum absolute error occur at the end point,. The Taylor polynomials for

of degree and at, , , , obtained as

(27)

Figure 4 illustrates that the difference between and is invisible on this scale. Figure 5 indicates the graph of the error over for the Taylor approximation. It is observed that the largest absolute error occur at the end point,. Figure 6" target="_self"> Figure 6 describes that the

difference between and is also invisible on this scale. Figure 7 explains the graph of the error over for the Taylor approximation. The maximum absolute error occur at the end point,. It is observed that the increase in the degree of Taylor polynomial increases the maximum absolute error.

In this part we discuss the graphs for the variation of the horizontal velocity profiles and shear stress at the wall with distance from the wall for different values of Reynolds number, local non-Newtonian parameter, fluid parameter, homotopy parameter and time.

Figures 8 and 9 describe the variation of the horizontal velocity profiles with for several values of by keeping, , and fixed. Figure 8 shows that when there is mass injection at

the bottom wall, with increase in fluid parameter, horizontal velocity profiles shows decreasing trend. Figure 9 shows that when there is mass suction at the top wall, with increase in, increases at all points. Figures 10 and 11 indicate the variation of the horizontal velocity profiles with for several values of by keeping, , and fixed. Figure 10 shows that when there is mass injection at the bottom wall, with increase in fluid parameter, horizontal velocity profiles increases at all points. Figure 11 shows that when there is mass suction at the top wall, with increase in, increases in magnitude but have negative values, an inverted behavior is observed, which is consistent with what we expected. Figures 12 and 13 illustrate the variation of the horizontal velocity profiles with for several values of time, for fixed values of, , and. Figures 12 and 13 are plotted for positive value of. Figure 12

shows that for mass injection at the bottom wall, with increase in, horizontal velocity profiles shows increasing trend in magnitude but have negative values. From Figure 13 it is clear that for mass

suction at the top wall, with increase in, increases at all points and the reverse behavior is observed. Figures 14 and 15 describe the variation of the horizontal velocity profiles with for several values of time, for fixed values of, , and. Figures 14 and 15 are plotted for negative value of. From Figure 14 it is observed that for mass injection at the bottom wall, with increase in, horizontal velocity profiles shows increasing trend in magnitude but have negative values. From Fig. 15 it is seen that for mass suction at the top wall, with increase in, increases at all points and have positive values, that is, a reverse trend is observed. From the comparison of the Figure 12 to 15 we observe that for positive and negative values of the variation of horizontal velocity profiles is same. The Figures 8 to 15 shows that mass transfer has a dominant effect on the horizontal velocity profiles. We observe from the graphs 8 to 15 that the

fluid material parameters and enhance the magnitude of the velocity profile. In Figures 8 to 15 it is observed that the behavior of suction is the reverse of the injection in all the cases, which is a confirmation for the validity of our results. Graphs from 8 to 15 are plotted for large values of the parameter, and, because for small values it is observed that the curves of different profiles overlaps and behavior is not clear, whether it is increasing or decreasing.

Figure 16 and 17 elucidate the variation of the shear stress at the wall with the parameter for several values of, for fixed values of, and. Figure 16 is for mass injection at the bottom wall and Figure 17 is for mass suction at the top wall. Figure 16 shows that with increase in, shear stress at the wall increases at all points for all values of and have positive values. Figure 17 shows that with increase in, shear stress at the wall