NUMERICAL METHODOLOGY

3.1 Governing equations

The conservative form of mass, momentum and energy conservation equations, using Einsteins summation convention over repeated indices, are then given by

Continuity equation:
(3.1)

Momentum equation:
* + (3.2)

According to the Stokess hypothesis which assumes that the bulk viscosity can be neglected, the shear-stress tensor for a Newtonian fluid is given by:

(3.3)

Energy equation:
[ ] ( is defined as: (3.5) ) (3.4)

where the viscous dissipation

3.2 Discretization approaches

To solve Navier-Stokes and Energy equations numerically, the flow domain needs to be discretized and the exact solution of the differential equations is replaced with values at discrete points in the domain. The discretizations are usually derived by three different approaches: 1. Finite Difference (FD) method 2. Finite Volume (FV) method 3. Finite Element (FE) method. In Finite Difference approach, the procedure for deriving finite difference equations consists of approximating the derivatives in the differential equation through a truncated Taylor series expansion. The grid point values of the variable are considered to constitute the solution, without an explicit reference as to how the variable varies between the grid points. The result is one algebraic equation per grid node, in which the variable value at that and a certain number of neighbor nodes appear as unknowns. On structured grids, the FD method is very simple and effective. It is especially easy to obtain higher-order schemes on regular grids. The disadvantage of FD methods is that the conservation is not enforced unless special care is taken. Also, the restriction to simple geometries is a significant disadvantage in complex flows. The Finite volume method uses the integral form of the conservation equations. The solution domain is subdivided into a finite number of contiguous control volumes (CVs), and the conservation equations are applied to each CV. For cell centered grid the centroid of each CV lies a computational node at which the variable values are to be calculated. In staggered grid the some variables are calculated at centroid of cell face. Interpolation is used to express variable values at the CV surface in terms of the nodal (CV-center) values. Surface and volume integrals are approximated using suitable quadrature formulae. An algebraic equation for each CV, in which a number of neighbor nodal values appear. The FV method can accommodate any type of grid, so it is suitable for complex geometries. The method is conservative by construction. The disadvantage of FV methods compared to FD schemes is that methods of order higher than second are more difficult to develop in 3D.

The Finite element method is similar to the FV method in many ways. The domain is broken into a set of discrete volumes or finite elements that are generally unstructured. In 2D, they are usually triangles or quadrilaterals, while in 3D tetrahedra or hexahedra are most often used. The distinguishing feature of FE methods is that the equations are multiplied by a weight function before they are integrated over the entire domain.

3.3 Computation of Turbulent flows

Most flows occurring in nature and engineering applications are turbulence. The boundary layers growing on aircraft wing, pipe flows, wakes of aircraft, car and ship are in turbulent motion. The study of turbulence clearly is an interdisciplinary activity, which has a very wide range of applications. Turbulence is an eddying motion which at the high Reynolds number usually prevailing has a wide spectrum of eddy size and a corresponding spectrum of fluctuating frequencies. The motion is always rotational and can be thought of a tangle of vortex elements whose vorticity vectors can be aligned in all directions and are highly unsteady. Due to the complexity of the governing equations of the flow, analytical solutions of these equations were restricted to very simple cases and, as a result an experimental approach was the only method to tackle this problem. The experimental approach is mainly based on similitude so that measurements made on one system, which is generally in the laboratory, can be used to describe the behavior of other systems. A large part of the fluid mechanics community are still involved in experimental analysis, but by the rapid growth of computer power and the reduction of their prices, experiments are becoming more and more expensive compared to the numerical approaches. Furthermore, in many situations like hypersonic flows or combustion problems with high temperature, measurements are very difficult or even impossible. In the numerical approach three main procedures are now in use for the computation of turbulent flows, Direct Numerical Simulation (DNS), Large-Eddy Simulation (LES), and a statistical approach based on the Reynolds Averaged Navier-Stokes (RANS) equations.

Figure 3.1: Energy cascading and different prediction methods The most accurate approach is Direct Numerical Simulation (DNS) and the NavierStokes equations are solved without averaging or approximation other than numerical discretization whose errors can be estimated and controlled. This method does not need any modeling and all the motions contained in the flow are resolved. The results of a DNS contain very detailed information about the flow at a large number of grid points. These results can be used to produce statistical information. DNS is used as a research tool for understanding the mechanisms of turbulence production, energy transfer, dissipation, noise production, drag reduction, and many other physical aspects of turbulence. In a direct numerical simulation all of the kinetic energy dissipation, which occurs on the smallest scales, is captured. The size of the grid must not be larger than a viscously determined scale, called the Kolmogorov scale, and the time step should be sufficiently small to resolve the period of the fastest fluctuation. If L is the characteristic length of the physical domain, the number of points in one direction should be of the order,
(3.6)

with the kinematic viscosity Kolmogorov length scale approximated as where

and the dissipation per unit mass

the can be

. The dissipation per unit mass

is a characteristic velocity of the flow. Finally the number

of points for a resolved DNS in three dimensions can be estimated as, ( ) ( )

(3.7)

For the high Reynolds numbers the number of points enormously increases and obviously DNS is restricted to relatively low Reynolds numbers. A method to overcome this problem is called Large-Eddy Simulation (LES) in which the low-frequency modes in space (large eddies) are directly simulated and the energy exchange with the high-frequency modes (small eddies), which are not explicitly simulated, is modeled using a sub-grid-scale model. The justification for such a treatment is that the larger eddies contains most of the energy and is clearly test case dependent whereas the smaller eddies are more universal and easier to model. To separate the large from the small scales, LES is based on the definition of a filtering operation. The Navier-Stokes equations are filtered and the effect of small scales appears through a sub-grid-scale (SGS) stress term. To close the system, an expression for the SGS stresses must be obtained. Similarly to DNS, LES provides a threedimensional, time dependent solution of the Navier-Stokes equations. Thus, it still requires fairly fine meshes. Among the objectives of LES are to study more complex configurations closer to those of engineering interest and at Reynolds numbers beyond the reach of DNS. In Reynolds Averaged Navier-Stokes (RANS) approach, all of the unsteadiness is averaged out. If the flow is steady, time averaging is used as a statistical averaging. While the small scales are somewhat universal, the large ones are affected very strongly by the boundary conditions. The complexity of turbulence makes it impossible for a single RANS model to represent all turbulent flows. Some adjustment of the constants is often required. In the present work RANS approach is used to model the turbulent flow instead of LES or DNS due to lack of computation power.

3.4 RANS Turbulence models

The chaotic behavior of the turbulent flow will experience fluctuations, which means that pressure and velocity will be defined as the mean value plus a fluctuation term; and

respectively. To handle this, the equations (continuity and momentum),

with additional terms due to fluctuations, are time averaged. The new equations are called Reynolds-Averaged Navier-Stokes (RANS) equations.

Continuity equation:
(3.8)

Momentum equation:
( ) * + ( ) (3.9)

where The terms ( which arise from the time averaging of the time dependent velocity

components are known as Reynolds stresses. It must be modeled in order to close the equation set. A common and relatively easy way to model these stresses is the Boussinesq approach which relates the Reynolds stresses to mean velocity gradients. ( ) ( ) (3.10)

The term

in the above equation is known as eddy viscosity or turbulent viscosity.

There are several different turbulence models within the RANS method that in different ways models the Reynolds stresses. Some models worth mentioning are Mixing length model, model, model and Reynolds stress model, which have none, two, two and seven additional

transport equations respectively, which needs to be solved in addition to the RANS equations. Naturally, all models have their advantages and disadvantages; the Mixing length model is not suitable for flows with separation and circulations, and the Reynolds stress model is the most complex one of these four and needs fairly large computer capability. Left are the two twoequation models and .

3.4.1
In the

Turbulence models
model the two transports equation are solved for the turbulent kinetic energy was initially is derived

and its dissipation rate . This original model also known as standard

proposed by Launder and Spalding (1972). For this model the transport equation for from the exact equation, but the transport for

was obtained using physical reasoning and is , but is not exact. The

therefore similar to the mathematically derived transport equation of turbulent kinetic energy following equations. *( ) +

, and its rate of dissipation , for this model are obtained by the

(3.11)

[(

where gradients,

(3.12) represents the generation of turbulent kinetic energy the arises due to mean velocity is the generation of turbulent kinetic energy that arises due to buoyancy, and

represents the fluctuating dilation in compressible turbulence that contributes to the overall dissipation rate. and are source terms for the continuous phase when modeling Eulerian

particle induced turbulence. , and are constants that have been determined experimentally and are taken to

have the following values; =1.44, =1.92, =0.09 . By

The available literature is not clear as to the specification of the coefficient default can be taken as constant everywhere or specified as follows { and

are turbulent Prandtl numbers for the turbulent kinetic energy and its

dissipation rate. These have also been derived experimentally and are defined as follows. ,

The turbulent (or eddy) viscosity at each point is related to the local values of turbulent kinetic energy and its dissipation rate by (3.11) where is constant and defined above. The term for the production of turbulent kinetic energy turbulence models studied and is defined as (3.12) is defined as is common in many of the

The generation of turbulent kinetic energy that arises due to buoyancy, follows

(3.13) where is the coefficient of thermal expansion, is the gravitationalvector and is the

turbulent Prandtl number. The dilation dissipation term, effects is defined as (3.14) where c is the speed of sound and Since the inception of the model, there have been countless attempts to improve it. model. This model contains a new is which accounts for turbulence from compressibility

A successful recent development is the realizable

transport equation for the turbulent dissipation rate. Also, a critical coefficient of the model,

expressed as a function of mean flow and turbulence properties, rather than assumed to be constant as in the standard model. This procedure lets the model satisfy certain mathematical constraints on the normal stresses consistent with the physics of turbulence. So the realizable model is substantially better than the standard model for many applications.