Turbulence in rough-wall boundary layers: universality issues

Wind tunnel measurements of turbulent boundary layers over three-dimensional rough surfaces have been carried out to determine the critical roughness height beyond which the roughness affects the turbulence characteristics of the entire boundary layer. Experiments were performed on three types of surfaces, consisting of an urban type surface with square random height elements, a diamond-pattern wire mesh and a sand-paper type grit. The measurements were carried out over a momentum thickness Reynolds number (Re _θ) range of 1,300–28,000 using two-component Laser Doppler anemometry (LDA) and hot-wire anemometry (HWA). A wide range of the ratio of roughness element height h to boundary layer thickness δ was covered (0.04 ￡ h/d ￡ 0.400.04 \leq h/\delta \leq 0.40). The results confirm that the mean profiles for all the surfaces collapse well in velocity defect form up to surprisingly large values of h/δ, perhaps as large as 0.2, but with a somewhat larger outer layer wake strength than for smooth-wall flows, as previously found. At lower h/δ, at least up to 0.15, the Reynolds stresses for all surfaces show good agreement throughout the boundary layer, collapsing with smooth-wall results outside the near-wall region. With increasing h/δ, however, the turbulence above the near-wall region is gradually modified until the entire flow is affected. Quadrant analysis confirms that changes in the rough-wall boundary layers certainly exist but are confined to the near-wall region at low h/δ; for h/δ beyond about 0.2 the quadrant events show that the structural changes extend throughout much of the boundary layer. Taken together, the data suggest that above h/δ ≈ 0.15, the details of the roughness have a weak effect on how quickly (with rising h/δ) the turbulence structure in the outer flow ceases to conform to the classical boundary layer behaviour. The present results provide support for Townsend’s wall similarity hypothesis at low h/δ and also suggest that a single critical roughness height beyond which it fails does not exist. For fully rough flows, the data also confirm that mean flow and turbulence quantities are essentially independent of Re _θ; all the Reynolds stresses match those of smooth-wall flows at very high Re _θ. Nonetheless, there is a noticeable increase in stress contributions from strong sweep events in the near-wall region, even at quite low h/δ.     

The turbulence structure of drag-reduced boundary layer flow

The structure of turbulence in a drag-reduced flat-plate boundary layer flow has been studied with particle image velocimetry (PIV). Drag reduction was achieved by injection of a concentrated polymer solution through a spanwise slot along the test wall at a location upstream of the PIV measurement station. Planes of velocity were measured parallel to the wall (x–z plane), for a total of 30 planes across the thickness of the boundary layer. For increasing drag reduction, we found a significant modification of the near-wall structure of turbulence with a coarsening of the low-speed velocity streaks and a reduction in the number and strength of near-wall vortical structures.     

An Attempt to Combine Large Eddy Simulation with the k-ε Model in a Channel-Flow Calculation

Large eddy simulation (LES) is combined with the Reynolds-averaged Navier–Stokes (RANS) equation in a turbulent channel-flow calculation. A one-equation subgrid-scale model is solved in a three-dimensional grid in the near-wall region whereas the standard k–ε model is solved in a one-dimensional grid in the outer region away from the wall. The two grid systems are overlapped to connect the two models smoothly. A turbulent channel flow is calculated at Reynolds numbers higher than typical LES and several statistical quantities are examined. The mean velocity profile is in good agreement with the logarithmic law. The profile of the turbulent kinetic energy in the near-wall region is smoothly connected with that of the turbulent energy for the k–ε model in the outer region. Turbulence statistics show that the solution in the near-wall region is as accurate as a usual LES. The present approach is different from wall modeling in LES that uses a RANS model near the wall. The former is not as efficient as the latter for calculating high-Reynolds-number flows. Nevertheless, the present method of combining the two models is expected to pave the way for constructing a unified turbulence model that is useful for many purposes including wall modeling. Received 11 June 1999 and accepted 15 December 2000     

Velocity-defect scaling for turbulent boundary layers with a range of relative roughness

Velocity profile measurements in zero pressure gradient, turbulent boundary layer flow were made on a smooth wall and on two types of rough walls with a wide range of roughness heights. The ratio of the boundary layer thickness (δ) to the roughness height (k) was 16≤δ/k≤110 in the present study, while the ratio of δ to the equivalent sand roughness height (k _s) ranged from 6≤δ/k _s≤91. The results show that the mean velocity profiles for all the test surfaces agree within experimental uncertainty in velocity-defect form in the overlap and outer layer when normalized by the friction velocity obtained using two different methods. The velocity-defect profiles also agree when normalized with the velocity scale proposed by Zagarola and Smits (J Fluid Mech 373:33–70, 1998). The results provide evidence that roughness effects on the mean flow are confined to the inner layer, and outer layer similarity of the mean velocity profile applies even for relatively large roughness.     

An experimental study of a two-dimensional plane turbulent wall jet

 Laser-Doppler measurements were conducted in a plane turbulent wall jet at a Reynolds number based on inlet velocity, Re ₀, of 9600. The initial development as well as the fully developed flow was studied. Special attention was given to the near-wall region, including the use of small measuring volumes and the application of specific near-wall data corrections, so that wall shear stresses were determined directly from the mean velocity gradient at the wall using only data below y ⁺=4. It was possible to resolve the inner peak in the streamwise turbulence intensity as well as the inner (negative) peak in the shear stress. Limiting values of (u′)⁺ and uv ⁺ were determined. Turbulence data from the outer region of the flow were compared to earlier hot wire measurements and large differences in the normal turbulence intensity and the shear stress were found. These differences can be attributed to high turbulence intensity effects on the hot-wires. Received: 17 October 1996 / Accepted: 8 December 1997     

An experimental contribution to near-wall measurements by means of a special laser Doppler anemometry technique

 A new experimental technique for the investigation of near-wall turbulence using laser Doppler anemometry is presented, which allows an accurate measurement of the flow field very close to the wall, with good resolution and a high data rate. Such a technique is tested in a fully developed turbulent flow (with Reynolds numbers between 4,300 and 67,000) by carrying out a careful statistical analysis of the streamwise and wall-normal velocity components within the near-wall region, at distances from the wall ranging from approximately y ⁺ = 1 to y ⁺ = 100. The velocity profiles, Reynolds stresses and higher-order moments of the two-dimensional boundary layer are presented. The results, which are in agreement with the most recent data in the literature, testify the validity of the proposed experimental solution. Moreover, the accuracy of the results allows the friction velocity to be calculated as the intercept at the wall of the best linear fit of the total stress profile; in this way, an unambiguous examination of the normalized statistics is possible. Received: 17 April 2001 / Accepted: 15 August 2001     

Adverse pressure gradient turbulent flows over rough walls

An experimental study of a fully developed turbulent channel flow and an adverse pressure gradient (APG) turbulent channel flow over smooth and rough walls has been performed using a particle image velocimetry (PIV) technique. The rough walls comprised two-dimensional square ribs of nominal height, k = 3 mm and pitch, p = 2k, 4k and 8k. It was observed that rib roughness enhanced the drag characteristics, and the degree of enhancement increased with increasing pitch. Similarly, rib roughness significantly increased the level of turbulence production, Reynolds stresses and wall-normal transport of turbulence kinetic energy and Reynolds shear stress well beyond the roughness sublayer. On the contrary, the distributions of the eddy viscosity, mixing length and streamwise transport of turbulence kinetic energy and Reynolds shear stress were reduced by wall roughness, especially in the outer layer. Adverse pressure gradient produced a further reduction in the mean velocity (in comparison to the results obtained in the parallel section) but increased the wall-normal extent across which the mean flow above the ribs is spatially inhomogeneous in the streamwise direction. APG also reinforced wall roughness in augmenting the equivalent sand grain roughness height. The combination of wall roughness and APG significantly increased turbulence production and Reynolds stresses except in the immediate vicinity of the rough walls. The transport velocities of the turbulence kinetic energy and Reynolds shear stress were also augmented by APG across most part of the rough-wall boundary layer. Further, APG enhanced the distributions of the eddy viscosity across most of the boundary layer but reduced the mixing length outside the roughness sublayer.     

A calibration scheme for quantitative concentration measurements using simultaneous PIV and PLIF

A new approach for calibration of planar laser-induced fluorescence (PLIF) measurements is presented. The calibration scheme is based on the fact that there is a constant concentration flux through each cross-section of a fluorescent plume in a given flow field and makes use of simultaneous measurements of particle image velocimetry (PIV) and PLIF. The following are the advantages of the current technique: (1) it is experimentally less demanding and (2) it does not require in situ calibration for generating the calibration curves. The technique can be implemented in many experimental setups (both in water and gaseous flows) provided the geometry of the time-averaged scalar field is known. Using the calibration scheme, an analysis is carried out on the measurements of concentration fields in grid turbulence to validate the proposed technique. To demonstrate the feasibility of the scheme, the distributed second-order moments (μ ₂), and concentration and velocity correlations ( á u^￠c^￠ ñ \left\langle {u^{\prime}c^{\prime}} \right\rangle and á v^￠c^￠ ñ \left\langle {v^{\prime}c^{\prime}} \right\rangle ) are computed. Good agreement is found with previous studies. In addition, a quantitative appraisal of a simple closure approximation of the moment-based transport equation is also presented using simultaneous PIV and PLIF.     

The Effect of Sweep-Angle Variation on the Turbulence Structure in a Separated,Three-Dimensional Flow

A three-dimensional separated flow behind a swept, backward-facing step is investigated by means of DNS for Re _H = C _∞ H/ν = 3000 with the purpose to identify changes in the statistical turbulence structure due to a variation of the sweep angle α from 0° up to 60°. With increasing sweep angle, the near-wall turbulence structure inside the separation bubble and downstream of reattachment changes due to the presence of an edge-parallel mean flow component W. Turbulence production due to the spanwise shear ∂W/∂y at the wall becomes significant and competes with the processes caused by impingement of the separated shear-layer. Changes due to a sweep angle variation can be interpreted in terms of two competing velocity scales which control the global budget of turbulent kinetic energy: the step-normal component U _∞ = C _∞cosα throughout the separated flow region and the velocity difference C _∞ across the entire shear-layer downstream of reattachment. As a consequence, the significance of history effects for the development into a two-dimensional boundary layer decreases with increasing sweep angle. For α ≥50°, near-wall streaks tend to form inside the separated flow region. Received 7 November 2000 and accepted 9 July 2002 Published online 3 December 2002 RID="*" ID="*" Part of this work was funded by the Deutsche Forschungsgemeinschaft within Sfb 557. Computer time was provided by the Konrad-Zuse Zentrum (ZIB), Berlin. Communicated by R.D. Moser     

Transitionally rough zero pressure gradient turbulent boundary layers

Near-wall measurements are performed to study the effects of surface roughness and viscous shear stresses on the transitionally rough regime (5 < k ⁺ < 70) of a zero pressure gradient turbulent boundary layer. The x-dependence is known from the eleven consecutive measurements in the streamwise direction, which allows for the computation of the streamwise gradients in the boundary layer equations. Thus, the skin friction is computed from the integrated boundary layer equation with errors of 3 and 5% for smooth and rough, respectively. It is found that roughness destroys the viscous layer near the wall, thus, reducing the contribution of the viscous stress in the wall region. As a result, the contribution in the wall shear stress due to form drag increases, while the viscous stress decreases. This yields Reynolds number invariance in the skin friction as k ⁺ increases into the fully rough regime. Furthermore, the roughness at the wall reduces the high peak of the streamwise component of the Reynolds stress in the near-wall region. However, for the Reynolds wall-normal and shear stress components, its contribution is not significantly altered for sand grain roughness.     

Investigation of turbulence structures in a drag-reduced turbulent channel flow with surfactant additive by stereoscopic particle image velocimetry

In the present study, we employed stereoscopic particle image velocimetry (PIV) to investigate the characteristics of turbulence structures in a drag-reduced turbulent channel flow with addition of surfactant. The tested drag-reducing fluid was a CTAC/NaSal/Water (CTAC: cetyltrimethyl ammonium chloride; NaSal: sodium salicylate) system at 25°C. The weight concentration of CTAC was 30 ppm. Stereoscopic PIV measurement was performed for a water flow (Re=1.1×10⁴) and a CTAC solution flow (Re=1.5×10⁴ with 54% drag reduction) in both the streamwise–spanwise and wall-normal-spanwise planes, respectively. The three-dimensionality of hairpin vortex structures in the near-wall region for wall-bounded turbulent flow was reproduced by conditionally averaging the stereoscopic two-dimensional-three-component velocity fields. A series of wall-normal vortex cores were found to align with the near-wall low-speed streaks with opposite vorticity signals at both sides of the streaks and with the vorticity decreased on average by about one order of magnitude in CTAC solution flow compared with water flow; the spanwise spacing between the near-wall low-speed streaks in the solution flow is increased by about 46%. The streamwise vorticity of the vortex cores appearing in the wall-normal-spanwise plane was also decreased by the use of drag-reducing surfactant additives.     

A New Low-Reynolds-Number One-Equation Model of Turbulence

In this study, we propose a new Low-Reynolds-Number (LRN)one-equation model, which is derived from an LRN two-equation(k-ε) model. The derivation of the transport equation, in principle, is based on the assumption that the turbulent structure parameter remains constant. However, the relation for the turbulent structure parameter a ₁(=|− |/k) is modified to account for near-wall turbulence. As a result, the present one-equation model contains a term which takes the near-wall limiting behavior explicitly into account. Thus, the present model provides the correct wall-limiting behavior of turbulence in the vicinity of the wall and can be applied to the analysis of heat transfer. The validity of the present model is tested in channel flows, boundary layer flows with and without pressure gradient, plane wall jet, and flow with separation and reattachment. The calculated results showed good agreement with the direct numerical simulation (DNS) and experimental data. This revised version was published online in July 2006 with corrections to the Cover Date.     

Finite-length effects on dynamical behavior of rod-like particles in wall-bounded turbulent flow

Combined Particle Image Velocimetry (PIV) and Particle Tracking Velocimetry (PTV) measurements have been performed in dilute suspensions of rod-like particles in wall turbulence. PIV results for the turbulence field in the water table flow apparatus compared favorably with data from Direct Numerical Simulations (DNS) of channel flow turbulence and the universality of near-wall turbulence justified comparisons with DNS of fiber-laden channel flow. In order to examine any shape effects on the dynamical behavior of elongated particles in wall-bounded turbulent flow, fibers with three different lengths but the same diameter were used. In the logarithmic part of the wall-layer, the translational fiber velocity was practically unaffected by the fiber length l. In the buffer layer, however, the fiber dynamics turned out to be severely constrained by the distance z to the wall. The short fibers accumulated preferentially in low-speed areas and adhered to the local fluid speed. The longer fibers (l/z > 1) exhibited a bi-modal probability distribution for the fiber velocity, which reflected an almost equal likelihood for a long fiber to reside in an ejection or in a sweep. It was also observed that in the buffer region, high-speed long fibers were almost randomly oriented whereas for all size cases the slowly moving fibers preferentially oriented in the streamwise direction. These phenomena have not been observed in DNS studies of fiber suspension flows and suggested l/z to be an essential parameter in a new generation of wall-collision models to be used in numerical studies.     

Effects of wall permeability on turbulence

In order to understand the effects of the wall permeability on turbulence near a porous wall, flow field measurements are carried out for turbulent flows in a channel with a porous bottom wall by a two-component particle image velocimetry (PIV) system. The porous media used are three kinds of foamed ceramics which have almost the same porosity (0.8) but different permeability. It is confirmed that the flow becomes more turbulent over the porous wall and tends to be turbulent even at the bulk Reynolds number of Re_b=1300 in the most permeable wall case tested. Corresponding to laminar to turbulent transition, the magnitude of the slip velocity on the porous wall is found to increase drastically in a narrow range of the Reynolds number. To discuss the effects of the wall roughness and the wall permeability, detailed discussions are made of zero-plane displacement and equivalent wall roughness for porous media. The results clearly indicate that the turbulence is induced by not only the wall roughness but the wall permeability. The measurements have also revealed that as Re_b or the wall permeability increases, the wall normal fluctuating velocity near the porous wall is enhanced due to the effects of the wall permeability. This leads to the increase of the turbulent shear stress resulting in higher friction factors of turbulence over porous walls.     

On near-wall hot-wire measurements

A specially constructed hot-wire probe was used to obtain very near-wall velocity measurements in both a fully developed turbulent channel flow and flat plate boundary layer flow. The near-wall hot-wire probe, having been calibrated in a specially constructed laminar flow calibration rig, was used to measure the mean streamwise velocity profile, distributions of streamwise and spanwise intensities of turbulence and turbulence kinetic energy k in the viscous sublayer and beyond; these distributions compare very favorably with available DNS results obtained for channel flow. While low Reynolds number effects were clearly evident for the channel flow, these effects are much less distinct for the boundary layer flow. By assuming the dissipating range of eddy sizes to be statistically isotropic and the validity of Taylor's hypothesis, the dissipation rate ɛ _iso in the very near-wall viscous sublayer region and beyond was determined for both the channel and boundary layer flows. It was found that if the convective velocity U _c in Taylor's hypothesis was assumed to be equal to the mean velocity Uˉ at the point of measurement, the value of (ɛ⁺ _iso)₁ thus obtained agrees well with that of (ɛ ⁺)_DNS for y ⁺ ≥ 80 for channel flow; this suggests the validity of assuming U _c=Uˉ and local isotropy for large values of y ⁺. However, if U _c was assumed to be 10.6u _τ , the value of (ɛ⁺ _iso)₂ thus obtained was found to compare reasonably well with the distribution of (ɛ⁺ _iso)_DNS for y ⁺≤ 15. Received: 31 May 1999/Accepted: 20 December 1999     

Dissipation rate estimation from PIV in zero-mean isotropic turbulence

Measuring the turbulent kinetic energy dissipation rate in an enclosed turbulence chamber that produces zero-mean flow is an experimental challenge. Traditional single-point dissipation rate measurement techniques are not applicable to flows with zero-mean velocity. Particle image velocimetry (PIV) affords calculation of the spatial derivative as well as the use of multi-point statistics to determine the dissipation rate. However, there is no consensus in the literature as to the best method to obtain dissipation rates from PIV measurements in such flows. We apply PIV in an enclosed zero-mean turbulent flow chamber and investigate five methods for dissipation rate estimation. We examine the influence of the PIV interrogation cell size on the performance of different dissipation rate estimation methods and evaluate correction factors that account for errors related to measurement uncertainty, finite spatial resolution, and low Reynolds number effects. We find the Re _λ corrected, second-order, longitudinal velocity structure function method to be the most robust method to estimate the dissipation rate in our zero-mean, gaseous flow system.     

Spatial and Temporal Characteristics of OH in Turbulent Opposed-Jet Double Flames

Simultaneous high repetition-rate, two-point hydroxyl (OH) time-series measurements with associated PLIF/PIV measurements are employed to investigate spatio–temporal scales and flame-velocity interactions in turbulent opposed jets sustaining methane-air double flames. For a fuel-side equivalence ratio, ϕ _B  = 1.2, a rich premixed flame exists on the fuel side while a diffusion flame exists on the air side of the stagnation plane. The bulk Reynolds number (Re) and strain rate (SR) can be adjusted to generate flames at ϕ _B  = 1.2 with both well separated and completely merged flame fronts. Simultaneous PLIF/PIV measurements highlight distinct spatial OH structures of the premixed and diffusive fronts corresponding to variations in the flow field. The self-propagating tendency of the rich premixed front causes large-scale wrinkling, thereby enhancing the OH contour length by 15% as compared to the diffusive front. Two-point OH time-series measurements are implemented to quantify both spatial and temporal fluctuations via study of radial length and time scales. In general, these integral length and time scales follow similar trends and reach a minimum at the axial location of peak [OH]. In comparison to merged double flames having higher Re and SR, greater OH fluctuations are observed in the rich-premixed front as compared to the diffusive front for a well separated double flame. Because of the developing turbulence, the OH length scales exhibit reduced axial gradients across the reaction zone for higher Re in comparison to lower Re. A stochastic time-series simulation, using a state relationship based on a joint mixture fraction and progress variable, is utilized to extract estimated scalar time scales from those of measured OH. The simulations indicate that the hydroxyl fluctuations in double flames are only twice those of the underlying conserved scalar. “Turbulent Opposed-Jet Double Flames” is submitted for consideration as a full length article to Flow Turbulence and Combustion.     

Comparison of near-wall treatment methods for high Reynolds number backward-facing step flow

A comparison of near-wall treatment methods using different turbulence models for flow over a backward-facing step is presented. A Reynolds number (Re) of about 38,000 (U _∞ = 44.2 m/s), based on the step height and the mean stream velocity, was considered. An appropriate near-wall treatment method is critical to the choice of turbulence model used to predict wall-bounded flow. Predictions were obtained by applying standard wall functions, non-equilibrium wall functions and a two-layer model with six different turbulence models. These results were compared with data by Driver and Seegmiller (“Backward-facing step with inclined opposite wall—experiments by driver and seegmiller”, 1985a, http://cfd.me.umist.ac.uk/ercoftac [2003, Jan 31]). Non-equilibrium wall functions with modified k ? ? models predicted the closest reattachment length. However, the two-layer model gave results more representative of the entire flow pattern. The predictions show that a proper combination of turbulence models and near-wall treatment methods give reliable results.     

Anisotropy measurements in the boundary layer over a flat plate with suction

Laser Doppler velocity measurements are carried out in a turbulent boundary layer subjected to concentrated wall suction (through a porous strip). The measurements are taken over a longitudinal distance of 9× the incoming boundary layer thickness ahead of the suction strip. The mean and rms velocity profiles are affected substantially by suction. Two-point measurements show that the streamwise and wall-normal autocorrelations of the streamwise velocity are reduced by suction. It is found that suction alters the redistribution of the turbulent kinetic energy k between its components. Relative to the no-suction case, the longitudinal Reynolds stress contributes more to k than the other two normal Reynolds stresses; in the outer region, its contribution is reduced which suggests structural changes in the boundary layer. This is observed in the anisotropy of the Reynolds stresses, which depart from the non-disturbed boundary layer. With suction, the anisotropy level in the near-wall region appears to be stronger than that of the undisturbed layer. It is argued that the mean shear induced by suction on the flow is responsible for the alteration of the anisotropy. The variation of the anisotropy of the layer will make the development of a turbulence model quite difficult for the flow behind suction. In that respect, a turbulence model will need to reproduce well the effects of suction on the boundary layer, if the model is to capture the effect of suction on the anisotropy of the Reynolds stresses.     

Experimental and Computational Investigations of Flow and Mixing in a Single-Annular Combustor Configuration

A complementary experimental and computational study of the flow and mixing in a single annular gas turbine combustor has been carried out. The object of the investigation is a generic mixing chamber model, representing an unfolded segment of a simplified Rich-Quick-Lean (RQL) combustion chamber operating under isothermal, non-reacting conditions at ambient pressure. Two configurations without and with secondary air injection were considered. To provide an appropriate reference database several planar optical measurement techniques (time-resolved flow visualisation, PIV, QLS) were used. The PIV measurements have been performed providing profiles of all velocity and Reynolds-stress components at selected locations within the combustor. Application of a two-layer hybrid LES/RANS (HLR) method coupling a near-wall k − ε RANS model with conventional LES in the core flow was the focus of the computational work. In addition to the direct comparison with the experimental results, the HLR performance is comparatively assessed with the results obtained by using conventional LES using the same (coarser) grid as HLR and two eddy-viscosity-based RANS models. The HLR model reproduced all important flow features, in particular with regard to the penetrating behaviour of the secondary air jets, their interaction with the swirled main flow, swirl-induced free recirculation zone evolution and associated precessing-vortex core phenomenon in good agreement with experimental findings.