Scribd Logo
Upload

Welcome to Scribd!

  • 22 views

    Turbulence and CFD: Beyond Modeling

    Uploaded by

    angusyoung1
    The document discusses limitations of the popular k-ε turbulence model and alternative approaches to modeling turbulence, including low Reynolds number k-ε models, k-ω models, Reynolds stress transport models, large eddy simulation, and direct numerical simulation. Reynolds stress transport models directly solve transport equations for Reynolds stresses but require closure models, while large eddy simulation resolves large scales and models small scales. Direct numerical simulation resolves all scales but is currently limited to simple geometries and low Reynolds numbers.

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as PDF, TXT or read online from Scribd

    Turbulence and CFD: Beyond Modeling

    Uploaded by

    angusyoung1
    22 views2 pages
    The document discusses limitations of the popular k-ε turbulence model and alternative approaches to modeling turbulence, including low Reynolds number k-ε models, k-ω models, Reynolds stress transport models, large eddy simulation, and direct numerical simulation. Reynolds stress transport models directly solve transport equations for Reynolds stresses but require closure models, while large eddy simulation resolves large scales and models small scales. Direct numerical simulation resolves all scales but is currently limited to simple geometries and low Reynolds numbers.

    Original Description:

    turbolence cfd

    Original Title

    turbolence

    Copyright

    © Attribution Non-Commercial (BY-NC)

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    The document discusses limitations of the popular k-ε turbulence model and alternative approaches to modeling turbulence, including low Reynolds number k-ε models, k-ω models, Reynolds stress transport models, large eddy simulation, and direct numerical simulation. Reynolds stress transport models directly solve transport equations for Reynolds stresses but require closure models, while large eddy simulation resolves large scales and models small scales. Direct numerical simulation resolves all scales but is currently limited to simple geometries and low Reynolds numbers.

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    22 views2 pages

    Turbulence and CFD: Beyond Modeling

    Uploaded by

    angusyoung1
    The document discusses limitations of the popular k-ε turbulence model and alternative approaches to modeling turbulence, including low Reynolds number k-ε models, k-ω models, Reynolds stress transport models, large eddy simulation, and direct numerical simulation. Reynolds stress transport models directly solve transport equations for Reynolds stresses but require closure models, while large eddy simulation resolves large scales and models small scales. Direct numerical simulation resolves all scales but is currently limited to simple geometries and low Reynolds numbers.

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 2

    1

    Turbulence and CFD: Beyond k modeling __________________________________________ Although two-equation closures such as the k model are arguably the most popular turbulence model for industrial applications of CFD, they have significant limitations and/or deficiencies. These are summarized below, together with some alternative approaches: 1. The use of wall functions is convenient but fundamentally not sound for complex flows where the underlying assumptions are violated, e.g. for impinging flows. In the 1990s extensive effort was given to the development of so called low Reynolds number (LRN) models, which implement carefully designed damping functions to ensure the correct near-wall behaviour of important parameters such as the Reynolds shear stress, which behaves as uv ( y + ) 3 in the near-wall region. 2. The k equation is sometimes used together with other turbulence variables. For example, in the k closure, instead of the dissipation rate, one uses the quantity = / k , which has units of [1 / T ] , i.e. represents an inverse turbulence time scale. In form, this equation set is very similar to the k model, and uses an eddy viscosity model (EVM) relation to reconstruct the Reynolds stress, where the eddy viscosity is now given by t = C k / . The k closure has the remarkable and useful property of being applicable all the way to the wall. 3. Most two-equation closures still suffer from the inherent limitation of using an EVM relation for the Reynolds stress terms, e.g.
    U U j u i u j = t i + x j xi + 2 k ij 3

    Recall that this constitutive relation was modeled in a similar way to that of molecular transport, i.e. it is a gradient transport model. For complex flows, where there is more than one important strain-rate, this model fails. It is also deficient for flows with strong curvature and buoyancy. One alternative to EVM closures is an approach which solves the Reynolds Stress Transport Equations (RSTE). These equations are formidable, since the Reynolds stress is a second-order tensor quantity. In order to close these equations, many of the terms must be represented by simpler model relations. Nonetheless, the framework of the transport equations does allow the complex coupling between the Reynolds stress terms and the mean velocity field to be much better represented. These models are especially applicable to flows with significant anisotropy, including near-wall flows.

    A symbolic representation of the RSTE is as follows:


    u iu j + U j u i u j = Dij + Pij + ij ij t xj

    where
    Dij

    is the diffusion term is the production term is the redistribution term is the destruction/dissipation term

    Pij ij

    ij

    Low Reynolds number or near-wall formulations have also been developed for the RSTE. As well, algebraic models can be derived from these equations by neglecting the net transport. 4. The challenge of modeling the Reynolds stress becomes problematic for many complex flows. (Recall that the Reynolds stress represents the contribution of the turbulent motions to the mean transport.) Furthermore, in some cases one requires specific information about the large-scale turbulent motions, e.g. in applications with unsteady or separated flows. This has led to the development of Large Eddy Simulation (LES) methods which calculate the large-scale motions and model the smaller-scale motions. These represent very challenging simulations since the unsteady large-scale motions must be calculated in threedimensions in a time-accurate manner. On the other hand, the expectation is that the residual stress (or subgrid-scale) terms can be adequately represented by simpler models, since the small-scale motions are more isotropic. In LES the large-scale variable is referred to as the filtered velocity field, U i . LES typically uses different calculation methods than do RANS models, although many CFD codes try to include this option. 5. Currently, the most accurate way to predict a turbulent flow is to use a Direct Numerical Simulation (DNS) in which all scales of motion are resolved by an unsteady, three-dimensional calculation. Examples of this approach include the pipe flow simulations of Wu and Moin (2008) (J Fluid Mech, Vol 608, pp. 81112) and the backward-facing step study used as a benchmark for the project. Even with massive computing resources, these types of simulations can only consider flows with simple geometry and low Reynolds number. From an engineering viewpoint, they are not practical since they discard the vast majority of the information being calculated. However, they presently have been used as refined experiments for exploring flow physics and calibrating turbulence models.

    You might also like