Week 9 - FVM Literature Review

Introduction

Numerical methods play a vital role in solving partial differential equations (PDEs) that arise in engineering and physical sciences, especially when analytical solutions are not feasible. Among the most widely used techniques are the Finite Difference Method (FDM) and the Finite Volume Method (FVM). These methods provide systematic approaches to approximate solutions of PDEs by discretizing the spatial domain.

The Finite Difference Method is based on approximating the derivatives in a differential equation using difference quotients at discrete grid points. It is relatively simple to implement and is particularly effective on structured, Cartesian grids. FDM directly applies the differential form of the governing equations and is best suited for problems with simple geometries and boundary conditions.

In contrast, the Finite Volume Method stems from the integral form of the conservation laws. It divides the domain into a finite number of control volumes and ensures the conservation of fluxes across the control volume boundaries. FVM is inherently conservative and can handle complex geometries and unstructured meshes more naturally, making it ideal for applications in fluid dynamics, heat transfer, and related fields.

This report explores both methods using a simple example—the 1D steady-state heat conduction equation—and highlights the key differences in their formulation and application. It also discusses the importance of interpolation schemes and flux limiters within the FVM framework, especially for accurately capturing gradients and discontinuities in field variables.

Finite Volume Method (FVM)

The Finite Volume Method solves conservation equations in their integral form over discrete control volumes.

Governing Equation (1D steady-state heat conduction):

d/dx(k*dT/dx)=0

Assume constant thermal conductivity kk, it simplifies to:

d²T/dx²=0

FVM Discretization:

Divide the domain into control volumes (CVs), where temperature is stored at cell centers. Consider a uniform grid with spacing Δx and control volume centered at node P, with neighbors W (west) and E (east).

x--------------------x----------------------x

E P W

1. Integrate over control volume Δx\Delta x around point PP:

∫_xw^xEd/dx(k*dT/dx)dx=0

Using the divergence theorem, this becomes:

K*dT/dx∣_xE − k*dT/dx∣_xW=0

Assuming k is constant:

k(T_E−T_PΔx)−k(T_P−T_WΔx)=0

This is the FVM discretized equation for node P.

Finite Difference Method (FDM)

The Finite Difference Method directly discretizes the differential form of the PDE using finite differences.

Governing Equation:

d²T/dx²=0

FDM Discretization:

Using central differences at interior node ii, with uniform spacing Δx:

d²T/dx²≈T_i+1−2T_i+T_i−1 Δx²

So the discretized equation is:

T_i+1 − 2T_i +T_i − 1Δx²=0

This looks mathematically identical to the FVM discretization above only in this simple case. However, the underlying approach is different:

Feature	FDM	FVM
Basis	Differential form	Integral form
Control Volume	Not explicitly used	Central to formulation
Conservation	Not inherently conservative	Guaranteed conservation across control volumes
Grid Flexibility	Structured, uniform grids mostly	Works on structured and unstructured meshes
Applicability to Complex Geometries	Limited	High

Need for Interpolation Schemes in FVM

In FVM, fluxes at control volume faces need to be computed. However, variable values are stored at centers. So, interpolation is needed to estimate values at the face.

Example:

To compute (dT/dx)_e we need:

T_e ≈ (T_P+T_E)/2

Other options:

• Upwind scheme: uses upstream value
• QUICK: uses three-point interpolation for higher accuracy

Need for Flux Limiters in FVM

In problems involving advection or convection, using high-order interpolation (like QUICK) can cause non-physical oscillations near discontinuities (e.g., shocks).

Flux limiters combine first-order (stable but diffusive) and second-order (accurate but oscillatory) schemes to:

• Maintain solution stability
• Avoid numerical oscillations
• Improve resolution near gradients

These are commonly used in Total Variation Diminishing (TVD) schemes.

Summary

• FVM uses control volume integration ensuring conservation of physical quantities.
• FDM uses pointwise approximation of derivatives.
• For simple cases (like 1D linear conduction), both yield similar algebraic forms, but FVM’s integral approach makes it more suitable for complex, real-world problems.
• Interpolation schemes are vital in FVM to estimate face values.
• Flux limiters help balance accuracy and stability in convection-dominated flows.

Conclusion

The Finite Volume Method is a powerful and flexible technique suited for solving conservation laws, especially in CFD. It ensures conservation by design and is applicable to complex geometries. Interpolation schemes are essential to approximate face values of variables, while flux limiters play a crucial role in maintaining accuracy and stability when dealing with steep gradients or discontinuities. Compared to FDM, FVM is more robust, particularly for flows and transport problems in real-world engineering applications.