Engineering Turbulence Modelling Research Articles

LES of turbulent flow past a swept fence

The present study deals with the large-eddy simulation of the span-wise invariant turbulent flow past a swept fence at low Reynolds numbers. The swept fence geometry was introduced by McCluskey et al. [Eighth Symposium on Turbulent Shear Flows, 1991, p. 9.5.1] as a way of gathering information on the behaviour of near wall flows beneath a separation region in the presence of a significant cross flow. The configuration avoids both the lack of generality of typical statistically two-dimensional flows and the difficulties of fully three-dimensional flows. In the present case, large-eddy simulation is both a challenging and promising approach. On the one hand, as in most low-Reynolds number and recirculating flows, simple wall models cannot be applied [Engineering Turbulence Modelling and Experiments 2, Florence, Italy, p. 303] on the other hand, due to the relatively low Reynolds of the flow and the type of scaling suggested in Hardman and Hancock [Exp. Fluids 27 (2000) 653], accurate resolution of the near wall region can be achieved without incurring prohibitively high computational costs. A variant of the dynamic SGS model of Germano et al. [Phys. Fluids A 3 (1991) 1760] with a smooth and reliable numerical behaviour is also tested in this flow. The agreement with the experimental data of Hardman [Moderately three-dimensional separated and reattaching turbulent flow. Ph.D. thesis, University of Surrey, 1998] is found to be satisfactory.

Structure and scales in turbulence modeling

The enstrophy of the large-scale energy-containing turbulence is proposed as the second turbulence scale for use, in conjunction with the turbulence energy, in two-scale one-point engineering turbulence models. Its transport equation is developed in general and modeled for homogeneous turbulence in terms of the two scales and our new one-point structure tensors. The model produces the correct behavior of the scales for both two- and three-dimensional turbulence. Constants in the high Reynolds number model are evaluated only by reference to asymptotic analysis for decaying turbulence in stationary and rotating frames, and this model is then shown to provide an excellent prediction of homogeneous turbulent shear flow when used with the structure tensors for that flow. The low Reynolds number constant in the model is evaluated using the asymptotic decay rate for isotropic turbulence at zero Reynolds number, and numerical simulations of decay for intermediate Reynolds numbers are used to establish one remaining constant, the value of which does not affect high Reynolds number predictions.

Application of CFD to environmental flows

Abstract This paper is concerned with the major issues considered to be important for successful application of computational fluid dynamics to environmental flows. Among the issues of primary concern in the present paper are meshing and turbulence modeling. As regards meshing, we propose an approach that employs unstructured meshes in conjunction with solution-adaptive mesh refinement. Examples are presented to illustrate the efficacy of the unstructured mesh-based approach. The issue of turbulence modeling is discussed at length, inasmuch as turbulence modeling determines the fidelity of computational results for environmental applications. Emphasis is laid upon the capability of engineering turbulence models to capture the salient features of environmental flows such as atmospheric boundary layer over a smooth terrain and flows around bluff bodies.

On the prediction of riblet performance with engineering turbulence models

On the prediction of riblet performance with engineering turbulence models

Collaborative Testing of Turbulence Models

A review is given of an ongoing international project, in which data from experiments on, and simulations of, turbulent flows are distributed to developers of (time-averaged) engineering turbulence models. The predictions of each model are sent to the organizers and redistributed to all the modelers, plus some experimentalists and other experts (total approx. 120), for comment. The 'reaction time' of modelers has proved to be much longer than anticipated, partly because the comparisons with data have prompted many modelers to improve their models or numerics.