7
Finite Volume Method

7.1 Introduction

Finite volume method (FVM) is another widely used numerical technique [1,2]. The fundamental conservation property of the FVM makes it the preferred method compared to various existing methods viz. finite difference method (FDM), finite element method (FEM), etc. In this approach, similar to the known numerical methods like FDM (Chapter 5) or FEM (Chapter 6), the volumes (elements or cells) are evaluated at discrete places over a meshed geometry. Then, the involved volume integrals of respective differential equation containing divergence term are converted to surface integrals by using well‐known divergence theorem [3,4]. Further, the simulated differential equation gets transformed over differential volumes into discrete system of algebraic equations. In the final step, the system of algebraic equations is solved by standard methods to determine the dependent variables.

Aforementioned FVM has various other advantages in handling differential equations occurring in science and engineering problems. For instance, a key feature of the FVM is formulation of physical space (domain of the differential equation) on unregulated polygonal meshes. Another feature is that the FVM is quite easy to implement various boundary conditions in a noninvasive manner, because the involved unknown variables are calculated at the centroids of the volume elements, rather than at their boundary faces [5]. These characteristics of the FVM have made it suitable for the numerical simulations in variety of applications viz. fluid flow, heat and mass transfer, etc. [6,7]. As such, in this chapter a brief background of the FVM is explained with respect to a simple example problem.

In the next section, we present the discretization techniques of the FVM.

7.2 Discretization Techniques of FVM

In the FVM, the considered domain on which the conservation laws are applied is subdivided into a set of nonoverlapping cells [ 4 , 5 ]. The conservation laws are applied on each cell to regulate the variables at some discrete points named as nodes. Similar to the FEM [8], there is a considerable freedom in the choice of cells and nodes (i.e. nodes may be at various locations of the cells, such as cell‐centers, cell‐vertices, or mid‐sides). As such, the complete geometrical approach of the FEM can be used in the FVM. There exist different types of cells viz. triangular, quadrilateral, etc. In this regard, Figure 7.1 shows some typical examples of choices of nodes with the associated definition variables [ 1 , 2 ].

Image described by caption.

Figure 7.1 Different discretized grids: (a) structured quadrilateral, (b) structured triangular, and (c) unstructured triangular.

The following section presents the general solving procedure of a convection–diffusion problem using the FVM.

7.3 General Form of Finite Volume Method

The FVM depends on approximate solution of the integral form with respect to conservation equations 2,9,10]. In the FVM, the given domain of differential equation is divided into a set of nonoverlapping finite volumes and then the respective integrals of the conservation equations are evaluated by using nodal (function) values at computational nodes. For easy understanding of readers, we present the FVM procedure by solving the convection–diffusion problem. The control volume domain of the convection–diffusion problem is presented in Figure 7.2.

Schematic displaying an irregular shape labeled V with a smaller irregular shape inside labeled dS. dS has arrows labeled n and u.

Figure 7.2 Sketch of the control volume V and its bounding surface S.

The general form of convection–diffusion problem [ 9 ,11] is represented as

(7.1)equation

where

  • images = Transient term (ρ‐Density, φ‐Conservative form of all fluid flow),
  •  · (ρuφ) = Convection term (u‐velocity),
  •  · (D∇φ) = Diffusion term (D‐Diffusion coefficient) and Sφ = Source term.

Integrating Eq. (7.1) over the given domain V, we obtain

(7.2)equation

For a fixed domain, images and images commute each other [11] , so we have

(7.3)equation

Then, using the Gauss‐divergence theorem [12], we change the volume integrals of Eq. (7.2) into surface integrals as

(7.4)equation
(7.5)equation

where n‐is the outward unit normal of the surface S (Fig. 7.2 ).

Now substituting Eqs. (7.3)–(7.5) in Eq. ( 7.2 ), the required integral form for the FVM is obtained as

(7.6)equation

7.3.1 Solution Process Algorithm

This section discusses the steps required for solving the convection–diffusion problem using the FVM.

  • Step (i): Discretize the given domain as a set of finite volumes and define the computational nodes to evaluate the required variables.
  • Step (ii): Apply integral form of conservation law to each finite volume.
  • Step (iii): Estimate the surface and volume integrals using quadrature formulae at computational nodes.
  • Step (iv): Assemble all the algebraic equations of the finite volumes to obtain a system of algebraic equations.
  • Step (v): Solve the obtained system of algebraic equations to find unknown values of the variables at computational nodes.

In Section 7.4, we implement the above steps to solve steady‐state convection–diffusion problem.

7.4 One‐Dimensional Convection–Diffusion Problem

The steady‐state convection–diffusion equation 9, 11 ] can be derived from transport equation 7.1) by replacing transient term images as

(7.7)equation

Then, the grid generation for solving Eq. (7.7) is illustrated in Section 7.4.1.

7.4.1 Grid Generation

In the initial step of grid generation, divide the considered domain into equal parts of small domain and place nodal points at the midway in between each small domain as given in Figure 7.3. Then, using these nodal points, create control volume near the edge in such a way that the physical boundaries coincide with control volume boundaries [ 5 , 6 ].

Schematic displaying a rectangle with a circle intersected by a horizontal line with circles labeled A, W, P, E, and B. P is at the center of the rectangle. Above the rectangle are two-headed arrows labeled δxWP and δxPE.

Figure 7.3 Grid generation by dividing the domain into discrete control volumes.

Further, place a general nodal point P for the control volume on the space between A and B in Figure 7.3 . The boundaries of control volumes are placed in the middle, between the adjacent nodes. Then, each node is enclosed by a control volume or cell. The nodes to the west and east are identified by W and E, respectively. The west‐side face and east‐side face of control volume are referred to as w and e, respectively. The distances between WP, wp, pe, and PE are denoted by δxWP, δxwp, δxpe, and δxPE, respectively.

7.4.2 Solution Procedure of Convection–Diffusion Problem

In the absence of source Sφ from Eq. ( 7.7 ), the steady‐state convection–diffusion in one‐dimensional flow field [ 11 ,13] is given by

(7.8)equation

Then, by integrating Eq. (7.8) as images over control volume [ 11 , 13 ], we may obtain

(7.9)equation

where the gradient images from east to west is calculated with the help of nodal points (Figure 7.3 ) as

(7.10)equation

Now, by substituting Eq. (7.10) in Eq. (7.9), we may get

(7.11)equation

Assume Sw = Se = S, δxPE = δxWP = δx, F = ρu (convective mass flux per unit area), and images (diffusion conductance) in order to obtain the discretized equation 9, 11 ]. Further, by rearranging Eq. (7.11), we obtain

(7.12)equation

where Fe = (ρu)e and Fw = (ρu)w.

The central difference approximation with linear interpolation [ 9 , 13 ,14] has been used to compute the cell face values φe and φw in a uniform grid

(7.13)equation

On substituting Eq. (7.13) in Eq. (7.12), we obtain

(7.14)equation

Rearranging Eq. (7.14), we get

7.15equation

After simplifying Eq. (7.15), we obtain the discretized equation as

(7.16)equation

where

(7.17)equation
(7.18)equation
(7.19)equation

and Fe = FW = F ⇒ Fe − FW = 0.

Now, by using Eqs. (7.17) and (7.18), one can rewrite Eq. (7.19) as

(7.20)equation

Finally, the linear system (discretized equations) (7.16) can be handled using any standard method to evaluate the unknown cell face values φP, φW, and φE.

Exercise

  1. 1 Calculate the steady‐state temperature in the rod for 1D heat conduction equation. Ends are maintained at constant temperature of 100 and 500 °C with L = 0.5 m, respectively and thermal conductivity K = 1000 W/mK, cross‐sectional area A = 10 × 10−3 m2, L = 1 cm (Hint: Take five control volumes).
  2. 2 Calculate the steady‐state temperature in a large plate of thickness L = 2 cm, constant thermal conductivity K = 0.5 W/mK, and uniform heat generation q = 1000 kW/m3. The faces A and B are at temperatures 100 and 200 °C, respectively.

References

  1. 1 LeVeque, R.J. (2002). Finite Volume Methods for Hyperbolic Problems, vol. 31. Cambridge: Cambridge University Press.
  2. 2 Eymard, R., Gallouët, T., and Herbin, R. (2000). Finite volume methods. Handbook of Numerical Analysis 7: 713–1018.
  3. 3 Godlewski, E. and Raviart, P.A. (2013). Numerical Approximation of Hyperbolic Systems of Conservation Laws, vol. 118. New York: Springer Science & Business Media.
  4. 4 Raithby, G.D. and Chui, E.H. (1990). A finite‐volume method for predicting a radiant heat transfer in enclosures with participating media. Journal of Heat Transfer 112 (2): 415–423.
  5. 5 Liu, Z.L. (ed.) (2018). Finite volume method. In: Multiphysics in Porous Materials, 385–395. Cham: Springer.
  6. 6 Demirdžić, I. and Perić, M. (1990). Finite volume method for prediction of fluid flow in arbitrarily shaped domains with moving boundaries. International Journal for Numerical Methods in Fluids 10 (7): 771–790.
  7. 7 Mingham, C.G. and Causon, D.M. (1998). High‐resolution finite‐volume method for shallow water flows. Journal of Hydraulic Engineering 124 (6): 605–614.
  8. 8 Bathe, K.J. and Wilson, E.L. (1976). Numerical Methods in Finite Element Analysis. Upper Saddle River: Prentice Hall.
  9. 9 Lazarov, R.D., Mishev, I.D., and Vassilevski, P.S. (1996). Finite volume methods for convection‐diffusion problems. SIAM Journal on Numerical Analysis 33 (1): 31–55.
  10. 10 Fuhrmann, J., Ohlberger, M., and Rohde, C. (eds.) (2014). Finite Volumes for Complex Applications VII‐Elliptic, Parabolic and Hyperbolic Problems: FVCA 7, Berlin, June 2014, Springer Proceedings in Mathematics and Statistics, vol. 78. Cham: Springer.
  11. 11 Stynes, M. (1995). Finite volume methods for convection‐diffusion problems. Journal of Computational and Applied Mathematics 63 (1–3): 83–90.
  12. 12 Pfeffer, W.F. (2016). The Divergence Theorem and Sets of Finite Perimeter. Boca Raton: Chapman & Hall/CRC.
  13. 13 Shukla, A., Singh, A.K., and Singh, P. (2011). A comparative study of finite volume method and finite difference method for convection‐diffusion problem. American Journal of Computational and Applied Mathematics 1 (2): 67–73.
  14. 14 Jasak, H. (1996). Error analysis and estimation for the finite volume method with applications to fluid flows. PhD thesis. Imperial College London.
  15. 15 Samarskii, A.A. (2001). The Theory of Difference Schemes, Monographs and Textbooks in Pure and Applied Mathematics, vol. 240. New York: Marcel Dekker.
..................Content has been hidden....................

You can't read the all page of ebook, please click here login for view all page.
Reset
3.147.48.244