Search

We present an effort to model the development and the control of the vortex breakdown phenomenon on a delta wing. The pair of the vortices formed on the suction side of a delta wing is the major contributor to the lift generation. As the angle of attack increases, these vortices become more robust, having high vorticity values. The critical point of a delta wing operation is the moment when these vortices, after a certain angle of attack, are detached from the wing surface and wing stall occurs. In order to delay or control the vortex breakdown mechanism, various techniques have been developed. In the present work, the technique based on the use of jet-flaps is numerically investigated with computational fluid dynamics by adopting two eddy-viscosity turbulence models. The computational results are compared with the experimental data of Shih and Ding (1996) . It is shown that between the two turbulence models, the more advanced one, which adopts a non-linear constitutive expression for the Reynolds-stresses, is capable to capture the vortex breakdown location for a variety of jet exit angles. The performance assessment of the models is followed by the investigation of the effect of the jet-flap on the lift and drag coefficients.

The turbulent kinetic energy budget in the wake generated by a high lift, low-pressure two-dimensional blade cascade of the T106 profile was investigated experimentally using hot-wire anemometry. The purpose of this study is to examine the transport mechanism of the turbulent kinetic energy and provide validation data for turbulence modeling. Point measurements were conducted on a high spatial resolution, two-dimensional grid that allowed precise derivative calculations. Positioning of the probe was achieved using a high accuracy traversing mechanism. The turbulent kinetic energy (TKE) convection, production, viscous diffusion and turbulent diffusion were all obtained directly from experimental measurements. Dissipation and pressure diffusion were calculated indirectly using techniques presented and validated by previous investigators. Results for all terms of the turbulent kinetic energy budget are presented and discussed in detail in the present work.

We present an effort to model the separation-induced transition on a flat plate with a semi-circular leading edge, using a cubic non-linear eddy-viscosity model combined with the laminar kinetic energy. A non-linear model, compared to a linear one, has the advantage to resolve the anisotropic behavior of the Reynolds-stresses in the near-wall region and it provides a more accurate expression for the generation of turbulence in the transport equation of the turbulence kinetic energy. Although in its original formulation the model is not able to accurately predict the separation-induced transition, the inclusion of the laminar kinetic energy increases its accuracy. The adoption of the laminar kinetic energy by the non-linear model is presented in detail, together with some additional modifications required for the adaption of the laminar kinetic energy into the basic concepts of the non-linear eddy-viscosity model. The computational results using the proposed combined model are shown together with the ones obtained using an isotropic linear eddy-viscosity model, which adopts also the laminar kinetic energy concept and in comparison with the existing experimental data.

A new effort to model the flow in a 90° rectangular duct by adopting three low-Reynolds-number turbulence models, two eddy-viscosity models (a linear and a non-linear) and a Reynolds-stress model, is presented. The complex flow development is a challenge for the application of turbulence models in order to assess their capability to capture the secondary flow and the developing vortices due to curvature and strong pressure gradient effects. The numerical results show that both the non-linear eddy-viscosity and the Reynolds-stress models can provide good results, especially for the velocity distributions. The superiority of the Reynolds-stress model is shown primarily in the Reynolds-stress distributions, which have the best quality among the predictions from the other models. On the other hand, the main advantage of the non-linear model is its simplicity and the smaller needed CPU cost, compared to the Reynolds-stress model. Additionally, in some stations of the flow development, the non-linear model provides good velocity distributions. The linear model gives lower quality predictions for the Reynolds-stress distributions, although it is capable in providing quite satisfactory results for the velocity distributions.

In the present work further numerical predictions for the flow field through a specific type of a heat exchanger, which is planned to be used in the exhaust nozzle of aircraft engines. In order to model the flow field through the heat exchanger, a porous medium model is used based on a simple quadratic relation, which connects the pressure drop with the inlet air velocity in the external part of the heat exchanger. The aim of this work is to check the applicability of the quadratic law in a variety of velocity inlet conditions configured by different angles of attack. The check is performed with CFD and the results are compared with new available experimental data for these inlet conditions. A detailed qualitative analysis shows that although the quadratic law has been derived for a zero angle of attack, it performs very well for alternative non-zero angles. These observations are very helpful since this simple pressure drop law can be used for advanced computations where the whole system of the exhaust nozzle together with the heat exchangers can be modeled within a holistic approach.

An experimental and computational study for the flow development through a heat exchanger for aero-engine applications is presented. The heat exchanger consists of elliptic tubes in a U formation, the ends of which are attached to the collector pipe, which has a cylindrical cross section. In this way, two identical packages (named as matrices) are formed and located at a certain distance between them. The elliptic tubes are placed in a staggered arrangement. Detailed flow measurements using a 3-hole pitot-static probe were carried out on a 1:1 scale model of the heat exchanger in order to measure the pressure drop through the heat exchanger and the velocity distribution behind it. The flow through the heat exchanger was modeled with a computational fluid dynamics approach. The heat exchanger matrices were modeled using a porous medium assumption. The pressure drop through each element of the porous medium was linked to an effective local velocity. In order to check the validity of the computational modeling, the results were compared to the measured flow parameters such as pressure and velocity distributions. Two sets of modeling were performed assuming a laminar and a turbulent flow. The results showed that the laminar approach gave better results and this is supported by the corresponding Reynolds numbers, which indicated that the global flow field is transitional.

The effect of a negative free-stream mean-shear velocity distribution on the boundary layer development on a flat plate with a semi-circular leading edge is studied experimentally and computationally. The geometry is the same as in the T3L test case of the ERCOFTAC Special Interest Group on Transition. The existence of a negative shear is related to the transition of the boundary layer from laminar to turbulent through separation. The flow investigated here has the same general characteristics as the one presented in a recent work by the authors, Palikaras et al. [Int. J. Heat Fluid Flow 23 (2002) 455–470], where the boundary layer development has been studied under free-stream conditions of uniform and positive mean-shear velocity distributions. The negative shear flow in the core region of the wind tunnel has a value ∂ U / ∂ y=−27.7 s−1, which is the opposite to the case examined by Palikaras et al. [Int. J. Heat Fluid Flow 23 (2002) 455–470]. In the first part of the paper, a detailed description of the flow is given. The measured quantities are presented, discussed and compared with the computational analysis in order to obtain a complete picture of the investigated flow. For the computations, the non-linear k–e model of Craft et al. [Int. J. Heat Fluid Flow 17 (1996) 108–115] is used, and satisfactory predictions are reported. In the second part, a detailed comparison of the results with the cases of uniform and positive mean-shear velocity inlet distribution is carried out. In the case of negative mean velocity the separated boundary layer leads to a larger reverse flow region than the two other cases. A relation is observed between the location of the stagnation point at the leading edge and the presence or absence of shear. When mean shear is present, depending on the sign, there is a movement of the stagnation point away from the symmetry line of the flat plate and it is believed that this is the driving mechanism that affects the boundary layer development and the longitudinal size of the reverse flow region. This remark is supported by the observation that for all the three cases studied, the longitudinal RMS distribution above the reverse flow region and in the free-stream region has the same values.

The effect of the free-stream velocity profile on the transition from laminar to turbulent flow on a flat plate was studied experimentally and numerically and it is presented in this paper. The flows investigated are based on the T3L test case of the ERCOFTAC Special Investigation Group for transition. According to this test case, the boundary layer development on a flat plate with a semi-circular leading edge is examined by means of transition due to separation, under various free-stream conditions concerning the turbulence intensity and velocity magnitude. In the present work, two free-stream velocity distributions were studied. The first was a uniform velocity one and the second, with a mean shear velocity profile with positive gradient, ∂ U / ∂ y=27.7 s−1. Measurements using hot-wire anemometry were taken in two primary regions: far upstream of the flat plate to observe the velocity and turbulence distributions and near the flat plate to capture the boundary layer development and the transition phenomenon. The effect of the two free-stream velocity distributions was studied and it was shown that for both velocity distributions a recirculation region of the flow occurred near the flat plate wall that led to transition dominated by the boundary layer separation. For the positive velocity gradient the separation region was smaller compared to the case of uniform free-stream profile. Both cases were also studied computationally. Two widely used linear eddy-viscosity turbulence models, the k–e and the k–ω with specific low Reynolds formulations were applied and in addition, a non-linear eddy-viscosity based on the k–e model has been implemented. In general, all the k–e models gave satisfactory predictions for both flow cases regarding the predicted velocity distributions, while the k–ω model gave poor results. Concerning the longitudinal Reynolds stress distributions in the near-wall region, the non-linear k–e model gave the best predictions inside the separation zone but it over predicted the corresponding values beyond the reattachment point while beyond the separation the linear models predicted the longitudinal stresses in a more satisfactory way.