Discretization of Equations in CFD Solver: A Detailed Guide to First Order, Second Order, QUICK, and MUSCL Schemes

Discretization is an important step in solving fluid flow and heat transfer problems in CFD. It means breaking down continuous equations into small parts that a computer can solve. These equations describe how fluids move and how heat transfers in systems. In ANSYS Fluent, a popular CFD software, different discretization methods can be used. Each method changes how the equations are solved. This affects the accuracy and speed of the simulation.

There are many discretization methods, but in this article, we focus on four main types: First Order Upwind, Second Order Upwind, QUICK, and MUSCL. These methods have different strengths and weaknesses. Our goal is to explain these four methods clearly. We will also show how they work by solving a case study in ANSYS Fluent. You will see how each method changes the results and what to choose for your simulations.

Figure 1: Types of Discretization of Equations in ANSYS Fluent.

 

Basic Concepts of Discretization in ANSYS Fluent

Discretization means changing continuous equations into small parts that a computer can solve. This is needed because fluid flow and heat transfer are described by complex equations. Computers cannot solve these equations directly in their continuous form. Instead, the equations are written for small parts of the fluid domain. This process is called discretization.

 

Governing Equations Solved in ANSYS Fluent

ANSYS Fluent solves important physical equations that describe fluid flow and heat transfer. These include:

• Navier-Stokes equations: These equations describe how fluid moves. They show how velocity and pressure change in space and time.
• Energy equation: This equation explains how heat moves in the fluid.
• Turbulence models: These equations model complex, chaotic fluid flow behavior that happens in many real cases.

The general transport equation in ANSYS Fluent can be written as:

Where:

Symbol	Meaning
Nfaces	Number of faces enclosing the control volume
ρ	Density of the fluid
φ	Transported scalar (e.g., velocity, temperature)
v	Velocity vector
𝐴f⃗	Area vector of the face f
Γφ	Diffusion coefficient for φ
Sφ	Source term per unit volume
V	Volume of the control cell

This equation shows how a general quantity φ moves and changes inside each small part (control volume) of the fluid.

Figure 2: This figure shows how ANSYS Fluent divides the fluid domain into control volumes to apply the finite volume method for discretizing transport equations.

 

How ANSYS Fluent Applies the Finite Volume Method for Discretization

ANSYS Fluent uses the Finite Volume Method (FVM) to solve these equations. FVM divides the fluid domain into many small control volumes, each surrounding a mesh cell. The integral form of the governing equations is applied to each control volume.

• The method ensures conservation of mass, momentum, and energy in each control volume.
• Fluxes of quantities like mass and heat are calculated across the faces of each control volume.
• This converts the continuous equations into algebraic equations for each cell.
• The algebraic equations are then solved numerically.

This approach works well with complex geometries and unstructured meshes, which are common in real-life simulations.

 

Types of Discretization Schemes Available in ANSYS Fluent

In the Fluent solver window, you can select different discretization schemes. These schemes decide how the values at the faces of control volumes are calculated from the values at cell centers. The choice affects the accuracy and stability of the solution.

Table 1: Common discretization schemes in Fluent include.

Term	Available Schemes	Description
Gradient	Green-Gauss Cell-Based, Green-Gauss Node-Based, Least Squares	Methods to calculate gradients in the mesh
Pressure	Second Order, Standard, PRESTO!	Schemes for pressure interpolation
Momentum (Convection Terms)	First Order Upwind, Second Order Upwind, QUICK, MUSCL, Bounded Central Differencing	Methods for convective flux calculation

Each scheme has its own balance between accuracy, stability, and computational cost. Selecting the right scheme depends on the problem and the mesh quality.

 

Overview of the Four Discretization Methods in ANSYS Fluent

ANSYS Fluent offers several discretization methods to calculate the values of the transported variable φ at the faces of control volumes. These methods affect the accuracy and stability of the solution. Below, we explain the four main discretization schemes: First Order Upwind, Second Order Upwind, QUICK, and MUSCL. We use the variable φ introduced earlier to show how each scheme calculates the face value φ_f .

First-Order Upwind Scheme

In the First-Order Upwind scheme, the value of φ at the face φ_f is set equal to the value of φ at the upstream cell center. This means the face value simply copies the value from the cell before the face in the flow direction.

This method assumes the cell-center value represents the average value over the whole cell.

Pros:

• Simple and stable.
• Good for coarse meshes or when stability is more important than accuracy.

Cons:

• Only first-order accurate, meaning it can be diffusive and smooth out sharp gradients.
• Lower accuracy for complex flows.

Typical Use:

• Initial runs to get a stable solution.
• Problems where accuracy is less critical.

 

Second-Order Upwind Scheme

The Second-Order Upwind scheme improves accuracy by using a Taylor series expansion to estimate φ_f. It calculates φ_f from the cell center value φ and its gradient ∇φ in the upstream cell, considering the distance vector  from the cell center to the face:

Here, φ and ∇φ are the value and gradient at the upstream cell center. The gradient ∇φ is limited to prevent new maxima or minima, keeping the solution stable.

Pros:

• More accurate than first order.
• Captures gradients better.

Cons:

• Slightly less stable, especially on coarse or skewed meshes.
• Requires gradient calculation.

Typical Use:

• When accuracy is important.
• Most steady-state and transient simulations.

 

QUICK Scheme (Quadratic Upstream Interpolation for Convective Kinematics)

The QUICK scheme uses a weighted average of second-order upwind and central difference values to calculate. For a one-dimensional control volume with face e, the face value is:

Where:

• are values at neighboring cell centers.
• are distances between cell centers and face.
• is a blending factor (ANSYS Fluent uses a solution-dependent ).

Pros:

• Higher-order accuracy on structured meshes.
• Reduces numerical diffusion compared to second-order upwind.

Cons:

• Mainly for structured (hexahedral or quadrilateral) meshes.
• May revert to second-order upwind on unstructured parts.

Typical Use:

• Structured mesh simulations needing high accuracy.

Figure 3: One-Dimensional Control Volume

 

MUSCL Scheme (Monotone Upstream-Centered Schemes for Conservation Laws)

The third-order MUSCL scheme blends a central difference scheme and the second-order upwind scheme to calculate :

Where:

• is the central difference value.
•  is the second-order upwind value.
• is the blending factor.

Pros:

• Applicable to arbitrary meshes (structured or unstructured).
• Reduces numerical diffusion significantly.
• Most accurate of the four schemes.

Cons:

• Can produce undershoots or overshoots in flows with sharp discontinuities (e.g., shock waves).
• No gradient limiter currently in ANSYS Fluent MUSCL implementation.

Typical Use:

• Complex 3D flows where high accuracy is required.

Table 2: Summary Table of Discretization Schemes

Scheme	Face Value φf Calculation	Accuracy	Stability	Mesh Type	Notes
First Order Upwind	φf = φupstream cell	First order	Very stable	All	Diffusive, low accuracy
Second Order Upwind	φf,SOU = φ + ∇φ · r→	Second order	Stable with limiters	All	Good balance of accuracy/stability
QUICK	Weighted average of second order and central difference (see formula)	Higher order	Stable on structured meshes	Structured (hexahedral, quadrilateral)	Higher accuracy, limited to mesh type
MUSCL	Blend of central difference and second order upwind (see formula)	Third order	Moderate	All (structured and unstructured)	High accuracy, may cause overshoot

 

Case Study Setup: Turbulent Non-Premixed Methane Combustion with Swirling Air

This case models methane and air entering an annular combustion chamber as separate streams. Air is injected as both a swirling jet and a co-flow with axial velocity. Methane enters separately and reacts with air inside the axisymmetric chamber. Radiation heat transfer is also included.

Geometry and Mesh

The figure illustrates the axisymmetric geometry of the combustion chamber used in the simulation. It features three distinct gas inlets on the left side: the methane inlet, the swirling air inlet, and the co-flowing air inlet. These inlets supply the reactants separately into the chamber.

The axis of symmetry runs horizontally through the center of the chamber, indicating that the geometry and flow are symmetric around this line. At the right end of the chamber, the outlet boundary allows the combustion products and gases to exit the domain. The flow exits freely, simulating an open boundary condition.

Figure 4: Geometry of the annular combustion chamber showing separate inlets for swirling air, co-flow air, and methane fuel and Axis (for symmetry) and Outlet.

The computational grid consists of structured cells precisely aligned with the axisymmetric geometry to ensure accurate resolution of the flow features within the chamber. To enhance solution accuracy, the grid density is increased in the inlet regions, where complex flow interactions and steep gradients occur. The mesh comprises a total of 61,280 square cells, carefully distributed to balance computational cost and result fidelity.

Figure 5: Mesh showing fine resolution in reaction zones and near inlets.

Boundary Conditions

Table 3: Boundary Conditions

Boundary	Condition
Methane inlet	Velocity: 32.7 m/s, species: 100% methane
Swirling air inlet	Axial velocity with swirl: 38.2 m/s
Co-flow air inlet	Axial velocity: 20 m/s, species: air
Walls	Adiabatic
Outlet	Pressure outlet, atmospheric pressure

 

Fluent Solver Settings

In ANSYS Fluent, the solver was set to model this problem with the following key features:

• Solver Type: Pressure-based, steady-state.
• Energy: on
• Turbulence Model: Realizable .
• Combustion Model: Non-premixed combustion model activated, with species transport enabled.
• Radiation Model: Discrete Ordinate Model.
• Species : Non-Premixed Combustion
• Discretization Schemes: The four discretization methods (First Order, Second Order, QUICK, MUSCL) were tested for momentum, energy, Swirl Velocity, Turbulent Kinetic Energy, Turbulent Dissipation Rate, Discrete Ordinates, Mean Mixture Fraction and Mixture Fraction Variance in separate runs for comparison.

Figure 6: Viscous Model Settings for Turbulent Flow Simulation

 

5. Fluent Solver Window and Discretization Settings

In ANSYS Fluent, the solver window allows users to control how the governing equations are solved. Among these controls, the discretization settings are essential to define how the transported variable φ is approximated at the faces of control volumes.

 

Discretization Settings in the Fluent Solver Window

In the Solution Methods panel of Fluent, you can find the Spatial Discretization section. Here, some options like Gradient and Pressure are often fixed or have limited choices:

• Gradient: Usually set to methods like Least Squares Cell Based for calculating variable gradients.
• Pressure: Commonly set to Second Order for accurate pressure interpolation.

However, for other equations such as Momentum, Swirl Velocity, Turbulent Kinetic Energy, Turbulent Dissipation Rate, Energy, Discrete Ordinates, Mean Mixture Fraction, and Mixture Fraction Variance, Fluent allows you to choose from four main discretization schemes:

• First Order Upwind
• Second Order Upwind
• QUICK
• MUSCL (Monotone Upstream-Centered Scheme for Conservation Laws)

These settings affect how the transported variable φ  is calculated at each cell face, impacting the accuracy and stability of the simulation.

Visual Examples from the Fluent Solver Window

The attached images show how Fluent’s solver window looks when different discretization schemes are selected for the same simulation case:

• First Image: All equations use the Third-Order MUSCL scheme, representing the highest accuracy setting among the four.
• Second Image: All equations are set to First Order Upwind, the simplest and most stable but least accurate scheme.
• Third Image: The Second Order Upwind scheme is applied to all relevant equations, providing a balance between accuracy and stability.
• Fourth Image: shows settings for the QUICK scheme.

Figure 7:  Solution Methods and Spatial Discretization Settings for MUSCL, First order, Second order and Quick Algorithms in ANSYS Fluent

Consistent Settings for Fair Comparison

To fairly compare the effects of these discretization methods on the simulation results, all other simulation parameters remain unchanged:

• The mesh is fixed and remains the same for all cases.
• Boundary conditions and physical models do not change.
• Only the discretization schemes for the transported variables φ are varied between the four modes.

Figure 8: The ANSYS Workbench window shows that the case study has been solved for 4 different modes of discretization of the equations.

Comparison of Simulation Results for Four Discretization Modes

In this section, we present and compare the simulation results obtained from the turbulent non-premixed methane combustion case solved using the four discretization methods: First Order Upwind, Second Order Upwind, QUICK, and MUSCL. The comparison is based on velocity and temperature contours, as well as numerical data extracted along two important lines in the domain.

Velocity and Temperature Contours

Figures of velocity and temperature contours visually represent how fluid flows and heat distributes inside the combustion chamber for each discretization mode. These contours provide an overview of the flow patterns, temperature fields, and combustion characteristics predicted by each scheme.

Figure 9: contour images of temperature for all four discretization modes.

Figure 10: contour images of velocity for all four discretization modes.

• The First Order Upwind contours show smoother gradients but less detail due to numerical diffusion.
• The Second Order Upwind contours reveal sharper flow structures and temperature gradients.
• The QUICK scheme results display even more precise capturing of flow and thermal features on the mesh.
• The MUSCL scheme provides the highest resolution in flow and temperature fields, showing finer details but may include minor oscillations near sharp gradients.

Lines for Data Extraction

Two lines are defined in the domain for detailed data analysis:

• Line 1: The axis of symmetry of the chamber. This line is important because it represents the centerline where species mixing and combustion strongly interact.
• Line 2: A line located 40 mm downstream from the gas inlets. This line is critical as it provides output data representing the state of flow and temperature after mixing and combustion develop.

Figure 11: Two defined lines to capture temperature and velocity data

Numerical Results Along Lines

The following data was extracted along these lines for each discretization mode:

1. Temperature on Line 1:
   The temperature profile along the axis of symmetry shows how well each discretization method predicts the combustion heat release and temperature distribution in the core flow. The Second Order Upwind, QUICK, and MUSCL schemes predict higher peak temperatures compared to the First Order Upwind, which tends to smooth the temperature field due to numerical diffusion and They differ only slightly from each other.

Figure 12: The temperature profile along the axis of symmetry (line 1) for First Order, Second Order, Quick and MUSCL.

1. Velocity on Line 2:
   The velocity values ​​at 40 mm downstream show how the flow develops after mixing. The MUSCL and QUICK schemes predict more accurate velocity values, indicating better resolution of the effects of swirling flow, while the First Order Upwind scheme underestimates the velocity due to the nature of the dispersion.

Figure 13:  The velocity profile along the line 2 for First Order, Second Order, Quick and MUSCL.

Temperature on Line 2:
The temperature data along line 2 shows the thermal state of the gases leaving the reaction zone. Similar to line 1, more accurate temperature peaks are observed for the Second Order Upwind, QUICK, and MUSCL designs, indicating their improved accuracy in capturing the heat released from combustion.

Figure 14:  The temperature profile along the line 2 for First Order, Second Order, Quick and MUSCL.

To summarize the differences, average values of temperature and velocity along Line 2 were calculated for each discretization mode:

Figure 15: Average Values on Line 2 for Temperature and velocity in different modes.

7. Conclusion

In this study, we investigated four main discretization schemes available in ANSYS Fluent—First Order Upwind, Second Order Upwind, QUICK, and MUSCL—and their effects on simulating turbulent non-premixed methane combustion with swirling air.

• First Order Upwind is very stable but introduces significant numerical diffusion, resulting in smoother but less accurate velocity and temperature predictions.
• Second Order Upwind improves accuracy by better capturing gradients, offering a good balance between accuracy and stability.
• QUICK provides higher-order accuracy on structured meshes, reducing numerical diffusion and better resolving flow and thermal features.
• MUSCL achieves the highest accuracy and detail by blending central differencing and upwind schemes, but may require careful handling to avoid oscillations near sharp gradients.

A key best practice is to maintain consistent mesh quality and boundary conditions when comparing discretization methods to ensure meaningful results.

Recommendations for Choosing Discretization Schemes in ANSYS Fluent

• For initial simulations or problems where numerical stability is critical, First Order Upwind is recommended.
• For general engineering applications requiring a balance between accuracy and stability, Second Order Upwind is suitable.
• When working with structured meshes and smooth flow fields, QUICK can provide improved accuracy.
• For complex 3D turbulent flows where high fidelity is needed, MUSCL is preferred, but users should be cautious of possible numerical oscillations.

Always perform mesh independence studies and validate results against experimental or benchmark data to select the most appropriate discretization scheme.

For readers interested in deeper understanding or advanced discretization techniques:

• The ANSYS Fluent Theory Guide offers comprehensive explanations of discretization schemes and their mathematical formulations.
• The Turbulent Non-Premixed Methane Combustion with Swirling Air case study presented here can be used as a step-by-step tutorial for practicing turbulent combustion simulations with swirling flow and radiative heat transfer.

In conclusion, the choice of discretization scheme in ANSYS Fluent strongly influences simulation accuracy, stability, and computational effort. Selecting the right scheme depends on the physics of the problem, mesh quality, and the desired accuracy. By understanding the strengths and limitations of each method, users can optimize their CFD simulations for reliable and efficient results.