Abstracts:The subgrid-scale (SGS) model is improved by using a scale-similarity model based on the analysis of the SGS force and SGS energy transfer around an elliptic Burgers vortex. Abe (2013) proposed an anisotropy-resolving SGS model in which the Bardina term of a scale-similarity model is mixed with an eddy viscosity model under a new concept wherein the Bardina term does not affect the SGS energy transfer although it affects the SGS force. By using the concept, we propose a scale-similarity model with the Clark term. The SGS energy transfer is determined by the scale-similarity term and not by the eddy viscosity term while the SGS force is improved by using the SGS kinetic energy (Abe, 2013). It is observed that the Clark term yields higher spatial correlation with the true distributions of the SGS force and SGS energy transfer around the elliptic vortex when compared to the Bardina term. The SGS model based on the Clark term exhibits good performance for turbulent channel flows with respect to $R{e}_{\tau }=180$ and 590 even in extremely coarse grid resolutions. Specifically, the SGS model with the SGS kinetic energy fairly improves the mean streamwise velocity profile.

Abstracts:In the present study we perform direct numerical simulations (DNSs) of fully-developed turbulent rectangular ducts with semi-cylindrical side-walls at Re τ, c  ≃ 180 with width-to-height ratios of 3 and 5. The friction Reynolds number Re τ, c is based on the centerplane friction velocity and the half-height of the duct. The results are compared with the corresponding duct cases with straight side-walls (Vinuesa et al., 2014), and also with spanwise-periodic channel and pipe flows. We focus on the influence of the semi-cylindrical side-walls on the mean cross-stream secondary flow and on further characterizing the mechanisms that produce it. The role of the secondary and primary Reynolds-shear stresses in the production of the secondary flow is analyzed by means of quadrant analysis and conditional averaging. Unexpectedly, the ducts with semi-cylindrical side-walls exhibit higher cross-flow rates and their secondary vortices relocate near the transition point between the straight and curved walls. This behavior is associated to the statistically preferential arrangement of sweeping events entering through the curved wall and ejections arising from the adjacent straight wall. Therefore, the configuration with minimum secondary flow corresponds to the duct with straight side-walls and sharp corners. Consequences on experimental facilities and comparisons between experiments and various numerical and theoretical models are discussed revealing the uniqueness of pipe flow.

Abstracts:The fine and large–scale structures of spatially developing uniformly sheared flow (USF) and USF distorted by a square–mesh grid were examined experimentally using one– and two–point, two–component hot–wire anemometry. In agreement with previous findings, it was confirmed that USF has a far downstream region, where the dissipation parameter C ε is approximately constant, and an intermediate region, where it scales with the local turbulent Reynolds number as $C_{\varepsilon }\propto {\text{Re}}_{\lambda }^{-0.6}$. The insertion of a grid across the USF resulted in the creation of multi-structure turbulence (a flow in which at least two kinds of dominant eddies exist and contribute significantly to the local averages at the energy-containing level) and a permanent reduction of the kinetic energy and the integral length scales of the turbulence. Near the grid the dissipation parameter scaled as $C_{\varepsilon }\propto {\text{Re}}_{\lambda }^{-1},$ as in decaying grid turbulence, despite the opposite evolution rates of Re λ in the two flows. The turbulence fine structure within this multi–structure region was unaffected significantly by the grid, whereas the large–scale anisotropy was markedly changed. Second–order structure functions, normalised by Kolmogorov scales, collapsed in the viscous sub–range , whereas within the inertial sub–range the effects of grid insertion were measurable.

Abstracts:Dimpled surfaces may enhance or decrease heat transfer compared to smooth surfaces. Here, the flow field of a round, axisymmetric air jet (diameter D) impinging on smooth and dimpled target surfaces (positioned at H/D ≈  5) was measured using particle image velocimetry (PIV). Reynolds numbers based on the jet exit velocity and D were, Re = 1,473, 6,322 and 12,438. Six different dimpled surfaces were investigated, subdivided in three pairs. In all cases, D/d >  2, where d is the dimple’s opening size and δ/d ≤  1.0, where δ is the dimple’s depth. The focus was on the primary and secondary vortex interaction along the developing wall jet that was studied using the vorticity and the directional swirling strength. The latter were used to extract the vortex cores and subsequently their numbers, areas and circulation. The results showed that despite significant changes to the surface morphology as a result of the dimples, primary - secondary vortex interaction along the developing wall jet was similar as for the smooth plate. Furthermore, the normalized strength of secondary vortices peaked at r/D ≈  2, but differences between the plates became negligible away from the stagnation point at r/D ≈  4. In addition, it was shown that outer layer self-similarity was attained for the two highest Re. The present results indicate that within the present geometrical constraints, the outer shear layer characteristics of the wall jet generated by an impinging round jet on dimpled surfaces is not fundamentally different from those on a smooth surface.

Abstracts:Direct numerical simulations of cube-roughened rough-to-smooth channel flows are performed with the objective of studying the response of turbulence statistics in the developing flow over smooth walls. Non-equilibrium effects persist and the global recovery is slow and incomplete by the streamwise exit of the computational domains, which is at about 10 channel half heights. The estimated recovery distance in the outer regions of the flow is on the order of 50 channel half heights, but different statistics have disparate relxataion rates. The turbulence structure swiftly relaxes to a ‘near’ equilibrium very close to the wall. Within this wall layer, due to a strong mean shear, turbulence statistics and instantaneous motions resemble their fully-smooth equivalents. However, the reversion is not complete because it is interrupted by large structures that persist from the upstream roughness. As the flow encounters the step change in roughness, it expands producing strong mean-advection effects, which prevent the canonical log-law region from being established. The expansion of the mean flow also results in an adverse pressure gradient across the channel. It recovers gradually, only becoming favourable near the exit of the computational domains.

Abstracts:The direct numerical simulation of a fully developed turbulent Couette–Poiseuille flow is performed to investigate the modification of turbulent statistics on the bottom and top walls compared to those in a pure Poiseuille flow. The streamwise mean velocity profile shows that an extended logarithmic layer for a Couette–Poiseuille flow is developed from each wall to the centerline. In addition, the turbulent intensities and Reynolds shear stress on the bottom wall are found to be larger than those in the Poiseuille flow, whereas it is reversed on the top wall due to reduction of the velocity shear. The quadrant analysis of the Reynolds shear stress reveals that large Q2 and Q4 event motions are continuously created throughout the entire flow near the centerline, leading to active momentum transport between the bottom and top walls for the C-type. Inspection of the pre-multiplied streamwise and spanwise energy spectra shows that distinct secondary outer peaks are created for all velocity components and a plateau, called the kx ⁻¹ region, is presented in the logarithmic region. Based on an analysis of the net force spectra, three spectral ranges in the wavelength space, corresponding to small- (λx /δ ≤ 1), large- (1 ≤ λx /δ ≤ 10) and very-large-scale (λx /δ ≥ 10) motions for a Couette–Poiseuille flow, are proposed, and very-large-scale structures are highly energetic and contribute more than half of the streamwise turbulent kinetic energy and Reynolds shear stress, where λx is the streamwise wavelength and δ is the channel half height.

Abstracts:Recently a new mechanistic model for pool and nucleate flow boiling was developed in our group. This model is based on the balance of forces acting on a bubble and considers the evaporation of the microlayer underneath the bubble, thermal diffusion around the cap of bubble due to the super-heated liquid and condensation due to the sub-cooled liquid. Compared to other models we particularly consider the temporal evolution of the microlayer underneath the bubble during the bubble growth by consideration of the dynamic contact angle and the dynamic bubble base expansion. This enhances, in our opinion, the model accuracy and generality. In this paper we further evaluate this model with experiments and direct numerical simulation (DNS) in order to prove the importance of dynamic contact angle and bubble base expansion.

Abstracts:The decay of homogeneous isotropic turbulence in a variable viscosity fluid with a viscosity ratio up to 15 is analyzed by means of highly resolved direct numerical simulations (DNS) at low Reynolds numbers. The question addressed by the present work is how quantities such as the kinetic energy and the associated dissipation rate, as well as the inter-scale transport mechanism of turbulence are changed by local fluctuations of the viscosity. The comparison is performed with respect to the decaying homogeneous isotropic turbulence with constant viscosity (CV), equal to the mean value of the variable viscosity (VV). From the one-point budget equation of the turbulent kinetic energy, it is shown that the mean dissipation rate is nearly unchanged by variable viscosity effects. This result is explained by a negative correlation between the local viscosity and the local velocity gradients. However, the dissipation is a highly fluctuating quantity with a strong level of intermittency. From a statistical analysis it is shown that turbulent flows with variable viscosity are characterized by an enhanced level of small-scale intermittency with respect to CV flow, which results in the presence of smaller length scales. The effect of variable viscosity on the turbulent cascade is analyzed by a budget equation for the velocity structure function. From DNS it is shown that viscosity gradients contribute to the inter-scale transport mechanism in the form of an inverse transport, where information propagates from the small scales to the large scales.

Abstracts:For the modeling of the interfacial structure characteristics, an experiment of vertical adiabatic air-water flow has been conducted under an atmosphere pressure condition. The experimental facility is composed of 5 × 5 rods arranged on a square pitch in a square casing with 9.5 mm outside rod diameter and 12.6 mm pitch, simulating a prototypic rod bundle in nuclear reactors. The miniaturized four-sensor conductivity probe has been developed to allow for the experimental measurement in the small flow channel. Extensive data are acquired for one-dimensional flow parameters including axial development of void fraction, interfacial area concentration and gas velocity. The effect of a prototypic spacer grids with mixing vanes on the flow structure has been discussed at various flow conditions based on the experimental data. Two competing effects, “swirling effect” and “bubble breaker effect”, are identified. The effect vanishes out within a short distance from the space grid. A drift-flux correlation is developed for predicting void fraction of adiabatic bubbly two-phase flow in a vertical rod bundle. Existing interfacial area correlations are tested and Hibiki-Ishii correlation (2002b) is recommended as most accurate correlation to predict the interfacial area concentration. It is expected that the newly obtained data in the 5 × 5 rod bundle is useful for developing the interfacial area transport equation and benchmarking a computational code.

Abstracts:The present paper represents a study on spurious noise generated in direct noise computation using finite volume methods. Different sources of spurious noise are examined as well as the mechanism of their generation. This investigation involves two test cases. The first one consists of a turbulence box convected by a uniform flow field. This case serves to identify spurious noise sources and qualitatively evaluate their relevance. The second test case involves a single side mirror mounted on a flat plate, which serves to quantify the level of spurious noise produced and its relevance compared to physical sound generated by the flow past the mirror. Both test cases are calculated using a compressible flow solver for low Mach-number flows utilised with an IDDES approach for turbulence modelling. A new acoustic damping model which damps out acoustic waves without affecting hydrodynamic turbulent fluctuations has been implemented. Furthermore, special emphasis on refinement interfaces is given. Since no measurements are available, the results of the direct noise computation are compared to the results of a method based on Kirchhoff integral.