Search results

This paper aims to compute flow and heat transfer through a straight, orthogonally rotating duct, with ribs along the leading and trailing walls, in a staggered arrangement and at an angle of 45° to the main flow direction.

Flow computations have been produced using a 3D non‐orthogonal flow solver, with two two‐layer models of turbulence (an effective‐viscosity model and a second‐moment closure), in which across the near‐wall regions the dissipation rate of turbulence is obtained from the wall distance. Flow comparisons have been carried out for a Reynolds number of 100,000 and for rotation numbers of 0 (stationary) and 0.1. Temperature comparisons have been obtained for a Reynolds number of 36,000, a Prandtl number of 5.9 (water) and rotation numbers of 0 and 0.2 and also at a Prandtl number of 0.7 (air) and a rotation number of 0.

It was found that both two‐layer models returned similar flow and thermal predictions which are also in close agreement with the flow and thermal measurements. The flow and thermal developments are found to be dominated by the rib‐induced secondary motion, which leads to strong span‐wise variations in the mean flow and the local Nusselt number and to a uniform distribution of turbulence intensities across the duct. Rotation causes the development of stronger secondary motion along the pressure side of the duct and also the transfer of the faster fluid to this side. The thermal predictions, especially those of the second‐moment closure, reproduce the levels and most of the local features of the measured Nusselt number, but over the second half of the rib interval over‐predict the local Nusselt number.

The work contributes to the understanding of the flow and thermal development in cooling passages of gas turbine blades, and to the validation of turbulence models that can be used for their prediction, at both effective viscosity and second‐moment closure levels.

This study is concerned with the computation of turbulent flow and heat transfer in U‐bends of strong curvature. Following the earlier studies within the authors' group on flows through round‐ended U‐bends, here attention is turned to flows through square‐ended U‐bends. Flows at two Reynolds numbers have been computed, one at 100,000 and the other at 36,000. In the heat transfer analysis, the Prandtl number was either 0.72 (air) or, in a further departure from our earlier studies, 5.9 (water). The turbulence modelling approaches examined, include a two‐layer and a low‐Re k‐ε model, a two‐layer and a low‐Re version of the basic differential stress model (DSM) and a more recently developed, realisable version of the differential stress model that is free of wall‐parameters. For the low‐Re effective viscosity model (EVM) and DSMs, an alternative, recently proposed length‐scale correction term, independent of wall distance has also been tested. Even the simplest model employed – two‐layer EVM – reproduces the mean flow development with reasonable accuracy, suggesting that the mean flow development is mainly influenced by mean pressure rather than the turbulence field. The heat transfer parameters, on the other hand, show that only the low‐Re DSMs produce reliable Nusselt number predictions for both Prandtl numbers examined.

To assess how effectively two‐layer and low‐Reynolds‐number models of turbulence, at effective viscosity and second‐moment closure level, can predict the flow and thermal development through orthogonally rotating U‐bends.

Heat and fluid flow computations through a square‐ended U‐bend that rotates about an axis normal to both the main flow direction and also the axis of curvature have been carried out. Two‐layer and low‐Reynolds‐number mathematical models of turbulence are used at effective viscosity (EVM) level and also at second‐moment‐closure (DSM) level. In the two‐layer models the dissipation rate of turbulence in the new‐wall regions is obtained from the wall distance, while in the low‐Re models the transport equation for the dissipation rate is extended right up to the walls. Moreover, two length‐scale correction terms to the dissipation rate of turbulence are used with the low‐Re models, and original Yap term and a differential form that does not require the wall distance (NYap). The resulting predictions are compared with available flow measurements at a Reynolds number of 100,000 and a rotation number (ΩD/U_bl) of 0.2 and also with heat transfer measurements at a Reynolds number of 36,000, rotation number of 0.2 and Prandtl number of 5.9 (water).

While the main flow features are well reproduced by all models, the development of the mean flow within the just after the bend in better reproduced by the low‐Re models. Turbulence levels within the rotation U‐bend are under‐predicted, but DSM models produce a more realistic distribution. Along the leading side all models over‐predict heat transfer levels just after the bend. Along the trailing side, the heat transfer predictions of the fully low‐Re DSM with the differential length‐scale correction term NYap are close to the measurements, with an average error of around 10 per cent, though at the bend exit it rises to 25 per cent. The introduction of a differential form of the length‐scale correction term to improve the heat transfer predictions of both low‐Re models.

The numerical models assumed that the flow remains steady and is not affected by large‐scale, low frequency fluctuations. Unsteady RANS computations or LES must also be tested in the future.

This work has expanded the range of complex turbulent flow over which the effectiveness of RANS models has been tested, to internal cooling flows simultaneously affected by orthogonal rotation and strong curvature.