Fill out the form to download

Required field
Required field
Not a valid email address
Required field
Required field
  • Set up your own cloud-native simulation in minutes.

  • Documentation

    CFD Numerics: Numerical Schemes

    Numerical schemes calculate the values for different terms like derivatives, e.g. gradient and interpolations of values from cell centers to nodes. A wide range of numerical schemes are available that provide flexibility and freedom. 

    KeywordCategory of mathematical terms
    interpolation schemespoint-to-point interpolations of values
    surface normal gradient schemescomponent of gradient normal to a cell face
    gradient schemesgradient \(∇\)
    divergence schemesdivergence \(∇ ∙\)
    Laplacian schemesLaplacian \(∇^2\)
    time differentiation schemesfirst and second-time derivatives \(∂ \over\ ∂t\) ,\(∂^2 \over\ ∂^2t\)
    Table 1: All available numerical schemes

    Time Differentiation

    The time differentiation schemes calculate the rate of change of the variables over time. When selecting a time scheme it must be noted that a problem designed for transient analysis will not necessarily run with steady-state and visa-versa.

    The table below shows all the possible types. Some may or may not be available depending upon analysis type.

    Eulerfirst-order, bounded, implicit
    local Eulerlocal time step, first order, bounded, implicit
    Crank Nicholson \(ψ\)second-order, bounded, implicit
    backwardsecond-order, implicit
    steady statedoes not solve for time derivatives
    Table 2: The selection of time-dependent behavior of the problem should be appropriately selected for the numerical schemes.

    The first order bounded, ‘Euler’ is the most robust while the second-order ones are more accurate. Backward is a second-order time-differentiation scheme that gives higher accuracy but may reduce stability.

    Interpolation Scheme

    The interpolation schemes define the terms that are interpolated values from cell centers to face centers. These are primarily used in the calculation of velocity to the face centers for the calculation of flux. There are numerous interpolation schemes but the default scheme in all the cases is the ‘Linear interpolation’ i.e a 2nd order interpolation scheme.

    Gradient Schemes

    The grad Schemes interpolate the values for the gradient terms in the differential equations. The discretization scheme available are listed as follows:

    Discretization schemeDescription
    Gauss <interpolationScheme>second-order, Gaussian integration
    least squaressecond-order, least squares
    fourth gradientfourth-order, least squares
    celllimited <gradScheme>cell limited version of one of the above schemes
    Table 3: All the discretization schemes available in SimScale

    ‘Gauss’ represents the standard finite volume discretization using the Gaussian integration that requires the interpolation from cell centers to face centers. The recommended types are the second-order accurate ‘Gauss Linear’ or ‘Least squares’ schemes. 

    For better stability and robustness, the ‘cellLimited’ versions of both can be used. The ‘Limiter coefficient’ of 1.0 means full boundedness/limiting of values while 0 means no limiting.


    The divergence schemes are used to calculate the convection term in the fluid dynamics differential equations \(∇ ∙(ρUU)\) :

    SchemeNumerical behavior
    linearsecond-order, unbounded
    skew linearsecond-order, (more) unbounded, skewness correction
    cubic correctedfourth-order, unbounded
    upwindfirst-order, bounded
    linear upwindfirst/second-order, bounded
    Table 4: The Gauss is the default choice for calculation and requires an interpolation method to be specified from this table.
    • Gauss upwind: first-order bounded, generally robust but compromises accuracy
    • Gauss linear: second-order, unbounded. Accurate but not robust
    • Gauss linear upwind: second-order, upwind-biased, unbounded, that requires discretization of the velocity gradient to be specified.
    • Gauss limited linear: a linear scheme that limits towards upwind in regions of rapidly changing gradient; requires a coefficient, where 1 is the strongest limiting, and shifting towards linear as the coefficient tends to 0.
    • Gauss linear upwind v \(∇u\): second-order, upwind, bounded, is a good choice for stable second-order linear scheme.

    Surface-normal Gradient

    A surface normal gradient is calculated at a cell face and is defined as the component normal to the face, of the gradient of values at the centers of the 2 cells.

    correctedexplicit non-orthogonal correction
    uncorrectedno non-orthogonal correction
    limited \(ψ\)limited non-orthogonal correction
    boundedbounded correction for positive scalars
    Table 5: A surface normal gradient is specified directly and is also used to evaluate a Laplacian term using Gaussian integration.

    For maintaining accuracy, an explicit non-orthogonal correction can be added to the orthogonal component, known as the corrected scheme. The correction increases in size as the non-orthogonality increases.
    Beyond 70°, the explicit correction can be so large that it can cause a solution to become unstable. The solution can be then stabilized by applying the limited scheme which requires a coefficient \(ψ\), 0 ≤ \(ψ\) ≤  1, where:

    • \(ψ\) = 0 : corresponds to uncorrected,
    • \(ψ\) = 0.333 : non-orthogonal correction ≤ 0.5 \( \times\) orthogonal part,
    • \(ψ\) = 0.5 : non-orthogonal correction ≤ orthogonal part,
    • \(ψ\) = 1 : corresponds to corrected.

    Usually, \(ψ\) is chosen to be 0.33 or 0.5, where:

    • 0.33 offers greater stability,
    • 0.5 greater accuracy.


    The typical Laplacian term is \(∇ ∙(ν∇U)\), which is the diffusion term in the momentum equations, that is calculated by the laplacianSchemes. Only the Gauss scheme is available for discretization and further requires a selection of both an interpolation scheme for the diffusion coefficient and a surface normal gradient scheme:

    Gauss <interpolationScheme> <snGradScheme>

    SchemeCategory of mathematical terms
    correctedunbounded, second-order, conservative
    uncorrectedbounded, first order, non-conservative
    limited \(ψ\)blend of corrected and uncorrected
    boundedfirst-order for bounded scalars
    Table 6: All Laplacian scheme options
    • Gauss linear corrected: is a 2nd order accurate with corrected gradients.
    • Gauss linear limited corrected 0.5: 2nd order, corrected, with limiter value 0.5

    Last updated: January 21st, 2022