0

Research Papers

J. Fluids Eng. 2013;136(1):011001-011001-6. doi:10.1115/1.4025608.

This work is concerned with the behavior of pulsatile flows over a backward-facing step geometry. The paper mainly focuses on the effects of the pulsation frequency on the vortex development of a 2:1 backward-facing step for mean Reynolds number of 100 and for 0.035 ≤ St ≤ 2.19. The dependence of the flow field on the Reynolds number (Re = 100 and 200) was also examined for a constant Strouhal number, St of 1. A literature survey was carried out and it was found that the pulsation modifies the behavior of the flow pattern compared to the steady flow. It was shown in the present work that the inlet pulsation generally leads to differences in the mean flow compared to the steady field although the inlet bulk velocity is the same due to energy redistribution of the large-scale vortices, which result in nonlinear effects. The particle-image velocimetry results show that the formation of coherent structures, dynamical shedding, and transport procedure are very sensitive to the level of pulsation frequencies. For low and moderate inlet frequencies, 0.4 ≤ St ≤ 1, strong vortices are formed and these vortices are periodically advected downstream in an alternate pattern. For very low inlet frequency, St = 0.035, stronger vortices are generated due to an extended formation time, however, the slow formation process causes the forming vortices to decay before shedding can happen. For high inlet frequencies, St ≥ 2.19, primary vortex is weak while no secondary vortex is formed. Flow downstream of the expansion recovers quickly. For Re = 200, the pattern of vortex formation is similar to Re = 100. However, the primary and secondary vortices decay more slowly and the vortices remain stronger for Re = 200. The strength and structure of the vortical regions depends highly on St, but Re effects are not negligible.

Commentary by Dr. Valentin Fuster

Research Papers: Flows in Complex Systems

J. Fluids Eng. 2013;136(1):011101-011101-8. doi:10.1115/1.4025466.

A ground vehicle traveling along a road is subject to unsteady crosswinds in a number of situations. In windy conditions, for example, the natural atmospheric wind can exhibit strong lateral gusts. Other situations, such as tunnel exits or overtaking induce sudden changes in crosswinds, as well. The interaction of this unsteady oncoming flow with the vehicle and the resulting aerodynamic forces and moments affect the vehicle stability and comfort. The objectives of the current study are to improve the understanding of flow physics of such transient flow and ultimately to develop measurement techniques to quantify the vehicle’s sensitivity to unsteady crosswind. A square back simplified car model is exposed to a forced oscillating yaw and results are compared to static measurements. Tests are conducted at Reynolds number Re = 3.7 × 105 and reduced frequencies ranging from 0.265 × 10−2 to 5.3 × 10−2. Unsteady side force and yawing moment measurements are associated with particle image velocimetry flow fields to interpret dynamic loads in link with flow topology evolution. Phase average force and moment measurements are found to exhibit a phase shift between static and dynamic tests that increases with oscillating frequency. Velocity fields reveal that the phase-shift seems to originate from the rear part of the car model. Moreover, lateral vortical structures appearing on the lee side from β = 15 deg increase this phase-shift and consequently appear to be favorable to the lateral stability of the vehicle.

Commentary by Dr. Valentin Fuster
J. Fluids Eng. 2013;136(1):011102-011102-10. doi:10.1115/1.4025365.

In the present paper, three-dimensional (3D) turbulent flow in the porous media formed by periodic arrays of particles is numerically investigated. 3D Navier–Stokes equations and a standard k-ε turbulence model with enhanced wall function are adopted to model the turbulent flow inside the pores. Both local and macroscopic turbulence characteristics for different particle types (cubic, spherical, and ellipsoidal particles) and array forms [simple cubic (SC) and body center cubic arrays (BCC)] with different pore Reynolds numbers and porosities are carefully examined. It is revealed that, in the structural arrays of particles, the effects of particle shape and array form would be remarkable. With the same Reynolds number and porosity, the magnitudes of turbulence kinetic energy and its dissipation rate for the simple cubic array of spheres (SC-S) would be higher than those for the other arrays. Furthermore, with a nonlinear fitting method, the macroscopic correlations for extra turbulence quantities $k∞$ and $ɛ∞$ in the structural arrays for different particle types and array forms are extracted. The forms of present correlations can fit well with those of Nakayama and Kuwahara's correlations [Nakayama and Kuwahara, 1999, “A Macroscopic Turbulence Model for Flow in Porous Media,” ASME J. Fluids Eng., 121(2), pp. 427–433], but some model constants would be lower.

Commentary by Dr. Valentin Fuster
J. Fluids Eng. 2013;136(1):011103-011103-11. doi:10.1115/1.4025460.

Previous computational and experimental studies that have demonstrated a method to generate vehicle maneuvering forces from a propulsor alone have been applied to a generic undersea vehicle. An open, preswirl propulsor was configured with an upstream stator row and downstream rotor. During normal operation, the upstream stator blades are all situated at the same pitch angle and preswirl the flow into the propulsor while generating a roll moment to counter the torque produced by the rotor. By varying the pitch angles of the stator blade about the circumference, it is possible to generate a mean stator side force that can be used to maneuver the vehicle. The stator wake axial velocity and swirl that is ingested into the rotor produces a counter-force by the rotor. Optimal design of the rotor minimizes the unsteady force and redirects the rotor force vector in an orthogonal direction to minimize the counter force. The viscous, 3D Reynolds-averaged Navier–Stokes (RANS) commercial code FLUENT® was used to predict the stator forces, velocity fields, and rotor response. Radiated noise was computed for the rotor separately and the entire geometry utilizing the Ffowcs Williams–Hawkings module available in FLUENT. Two separate geometries were studied—the first with a maximum stator blade row diameter contained within the body diameter and a second that was allowed to exceed the body diameter. Side force coefficients were computed for the two maneuvering propulsor configurations and compared with currently used control surface forces. Computations predicted that the maneuvering propulsor generated side forces equivalent to those produced by conventional control surfaces with side force coefficients on the order of 0.3. This translates to 50% larger forces than can be generated by conventional control surfaces on 21 in. unmanned undersea vehicles. Radiated noise calculations in air demonstrated that the total sound pressure levels produced by the maneuvering propulsor were on the order of 5 dB lower than the control fin test cases.

Commentary by Dr. Valentin Fuster
J. Fluids Eng. 2013;136(1):011104-011104-9. doi:10.1115/1.4025532.

An experimental study was conducted to get more insight into the flowing characteristics of single phase spiral flow in the horizontal pipe by the use of a laser Doppler velocimeter (LDV). Water was used as the working medium, and the spiral motion was produced by a vane. The vanes with different spiral angles and vane area were self-made. Influence of flow attenuation, average Reynolds number, spiral angle, and vane area on axial velocity distribution and tangential velocity distribution were studied. Turbulence intensity distribution was studied, and the spiral strength attenuation law was analyzed. The experimental results show that the vane is an efficient spiral device with low pressure drop, and it is used in pipeline, natural gas hydrate formation, and so on.

Commentary by Dr. Valentin Fuster
J. Fluids Eng. 2013;136(1):011105-011105-10. doi:10.1115/1.4025513.

The throat tap nozzle of the American Society of Mechanical Engineers performance test code (ASME PTC) 6 is widely used in engineering fields, and its discharge coefficient is normally estimated by an extrapolation in Reynolds number range higher than the order of 107. The purpose of this paper is to propose a new relation between the discharge coefficient of the throat tap nozzle and Reynolds number by a detailed analysis of the experimental data and the theoretical models, which can be applied to Reynolds numbers up to 1.5 × 107. The discharge coefficients are measured for several tap diameters in Reynolds numbers ranging from 2.4 × 105 to 1.4 × 107 using the high Reynolds number calibration rig of the National Metrology Institute of Japan (NMIJ). Experimental results show that the discharge coefficients depend on the tap diameter and the deviation between the experimental results and the reference curve of PTC 6 is 0.75% at maximum. New equations to estimate the discharge coefficient are developed based on the experimental results and the theoretical equations including the tap effects. The developed equations estimate the discharge coefficient of the present experimental data within 0.21%, and they are expected to estimate more accurately the discharge coefficient of the throat tap nozzle of PTC 6 than the reference curve of PTC 6.

Commentary by Dr. Valentin Fuster
J. Fluids Eng. 2013;136(1):011106-011106-7. doi:10.1115/1.4025646.

Most fish share a common geometry, a streamlined anterior body and a deep caudal fin, connected to each other at a tail-base neck, where the body almost shrinks to a point. This work attempts to explain the reason that fish exhibit this type of geometry. Assuming that the fish-like geometry is a result of evolution over millions of years, or, that bodies of modern-day fish have been optimized in some manner as a result of evolution, this work investigates the optimum geometry for a swimming object through existing mathematical optimization techniques to check whether the result obtained is the same as the naturally observed fish-like geometry. In this analysis, the work done by a swimming object is taken as the objective function of the optimization. It is found that a fish-like geometry is in fact obtained mathematically, provided that the appropriate constraints are imposed on the optimization process, which, in turn, provides some clues that explain the reason that fish have a fish-like geometry.

Commentary by Dr. Valentin Fuster
J. Fluids Eng. 2013;136(1):011107-011107-5. doi:10.1115/1.4025648.

The lateral and longitudinal spacing between individual turbines in a wind turbine array must be large enough to minimize the wake effects caused by an upstream turbine on those that lie downstream from it. Here, the flow downstream of a single wind turbine is examined by modeling its far-field development as a turbulent axisymmetric wake which is well described in the turbulence literature. In particular, the velocity defect profile in the wake is approximated by a Gaussian function in the radial coordinate. Scaling laws are used to derive closed-form solutions for the wake diameter, wake velocity defect, and wake power recovery as functions of downstream distance from the rotor. Our results show that at a downstream distance of 10 rotor diameters, the wake centerline velocity will recover to 77% of the free-stream value. It is also seen that the power within the wake recovers quickly for small downstream distances, but beyond about 10 rotor diameters, the rate of power recovery slows down. Implications for the optimal spacing of wind turbines are discussed based on these findings.

Commentary by Dr. Valentin Fuster

Research Papers: Fundamental Issues and Canonical Flows

J. Fluids Eng. 2013;136(1):011201-011201-13. doi:10.1115/1.4025363.

In order to study the flow behavior of multiple jets, numerical prediction of the three-dimensional domain of round jets from the nozzle edge up to the turbulent region is essential. The previous numerical studies on the round jet are limited to either two-dimensional investigation with Reynolds-averaged Navier–Stokes (RANS) models or three-dimensional prediction with higher turbulence models such as large eddy simulation (LES) or direct numerical simulation (DNS). The present study tries to evaluate different RANS turbulence models in the three-dimensional simulation of the whole domain of an isothermal, low Re (Re = 2125, 3461, and 4555), free, turbulent round jet. For this evaluation the simulation results from two two-equation (low Re $k-ɛ$ and low Re shear stress transport (SST) $k-ω$), a transition three-equation ($k-kl-ω$), and a transition four-equation (SST) eddy-viscosity turbulence models are compared with hot-wire anemometry measurements. Due to the importance of providing correct inlet boundary conditions, the inlet velocity profile, the turbulent kinetic energy ($k$), and its specific dissipation rate ($ω$) at the nozzle exit have been employed from an earlier verified numerical simulation. Two-equation RANS models with low Reynolds correction can predict the whole domain (initial, transition, and fully developed regions) of the round jet with prescribed inlet boundary conditions. The transition models could only reach to a good agreement with the measured mean axial velocities and its rms in the initial region. It worth mentioning that the round jet anomaly is still present in the turbulent region of the round jet predicted by the low Re $k-ɛ$. By comparing the $k$ and the $ω$ predicted by different turbulence models, the blending functions in the cross-diffusion term is found one of the reasons behind the more consistent prediction by the low Re SST $k-ω$.

Commentary by Dr. Valentin Fuster
J. Fluids Eng. 2013;136(1):011202-011202-9. doi:10.1115/1.4025509.

In this paper some aspects of the unsteady friction in pipe flow expressed by the convolution are analyzed. This additional term introduced into the motion equation involves the accelerations of fluid occurring in the past and a weighting function. The essence of such approach is to assume the appropriate form of weighting function. However, until now, no fully reliable formula for this function has been found. To avoid some inconveniences typical for the commonly used weighting functions, an alternative form of the convolution is proposed. Instead of a weighting function an impulse response function in a general form is introduced. This function, defined in the real time domain, having clear physical interpretation and some useful properties is not related to the usually assumed viscosity distribution over the pipe's cross section. The proposed approach involves two parameters. The convergence of the impulse response function, characterized by the flow memory, is determined by a parameter which can be related to the pressure wave frequency. The second parameter determines the magnitude of the unsteady friction force. The proposed alternative convolution approach was tested basing on the laboratory measurements for a water hammer event initiated by turbulent flow in pipes made of steel. Although the alternative convolution approach causes a very good damping of the pressure wave amplitude, it appears to be unable to ensure appropriate smoothing of the pressure heads. This is because it acts in the dynamic equation as a source/sink term. To ensure the required smoothing of the pressure wave the diffusive term was included into the dynamic equation.

Commentary by Dr. Valentin Fuster

Research Papers: Multiphase Flows

J. Fluids Eng. 2013;136(1):011301-011301-16. doi:10.1115/1.4025364.

Numerical simulations using an Eulerian two-fluid model were performed for spatially developing, two-dimensional, axisymmetric jets issued from a 30-mm-diameter circular nozzle. The nozzle was simulated separately for various flow conditions to get fully developed velocity profiles at its exit. The effect of interparticle collisions in the nozzle gives rise to solids pressure and viscosity, which are modeled using kinetic theory of granular flows (KTGF). The particle sizes are in the range of 30 $μm$ to 2 mm, and the particle loading is varied from 1 to 5. The fully developed velocity profiles are expressed by power law, $U=Uc(1-(r/R))N$. The exponent, N, is found to be 0.14 for gas phase, irrespective of particle sizes and particulate loadings. However, the solid-phase velocity varies significantly with the particle diameter. For particle sizes up to 200 $μm$, the exponent is 0.12. The center line velocity ($Uc$) of the solid phase decreases and, hence, the slip velocity increases as the particle size increases. For 1 mm and 2 mm size particles, the exponent is found to be 0.08 and 0.05, respectively. The developed velocity profiles of both the phases are used as the inlet velocities for the jet simulation. The modulations on the flow structures and turbulent characteristics of gas flow due to the solid particles with different particle sizes and loadings are investigated. The jet spreading and the decay of the centerline mean velocity are computed for all particle sizes and loadings considered under the present study. Additions of solid particles to the gas flow significantly modulate the gas turbulence in the nozzle as well as the jet flows. Fine particles suppress the turbulence, whereas coarse particles enhance it.

Commentary by Dr. Valentin Fuster

Technical Briefs

J. Fluids Eng. 2013;136(1):014501-014501-9. doi:10.1115/1.4025454.

This paper presents an automatic multiobjective hydrodynamic optimization strategy for pump–turbine impellers. In the strategy, the blade shape is parameterized based on the blade loading distribution using an inverse design method. An efficient response surface model relating the design parameters and the objective functions is obtained. Then, a multiobjective evolutionary algorithm is applied to the response surface functions to find a Pareto front for the final trade-off selection. The optimization strategy was used to redesign a scaled pump–turbine. Model tests were conducted to validate the final design and confirm the validity of the design strategy.

Commentary by Dr. Valentin Fuster
J. Fluids Eng. 2013;136(1):014502-014502-7. doi:10.1115/1.4025644.

This paper deals with a computational fluid dynamics (CFD) and experimental drag analysis on an isolated rotating wheel subsystem (including its accessories: tire, suspension, A-arms, and fender) of a motor tricycle vehicle with two wheels in front. The main goal of the present work is to study the effect of the fender on the wheel subsystem drag and its optimization. The Star CCM+ commercial code was used for the numerical simulations. Different flow conditions were simulated and some results were validated by comparison to wind tunnel experimental results. To perform drag optimization, several aerodynamic fender shapes were designed and simulated as part of the subsystem. A drastic drag reduction up to 30.6% compared to the original wheel subsystem was achieved through numerical simulations.

Commentary by Dr. Valentin Fuster
J. Fluids Eng. 2013;136(1):014503-014503-6. doi:10.1115/1.4025647.

Carbon dioxide is an attractive alternative to conventional refrigerants due to its low direct global warming effects. Unfortunately, CO2 and many alternative refrigerants have lower thermodynamic performance resulting in larger indirect emissions. The effective use of ejectors to recover part of the lost expansion work, which occurs in throttling devices, can close this performance gap and enable the use of CO2. In an ejector, the pressure of the motive fluid is converted into momentum through a choked converging-diverging nozzle, which then entrains and raises the energy of a lower-momentum suction flow. In a two-phase ejector, the motive nozzle flow is complicated by the nonequilibrium phase change affecting local sonic velocity and leading to various types of shockwaves, pseudo shocks, and expansion waves inside or outside the exit of the nozzle. Since the characteristics of the jet leaving the motive nozzle greatly affect the performance of the ejector, this paper focuses on the details of flow development and shockwave interaction within and just outside the nozzle. The analysis is based on a high-fidelity model that incorporates real-fluid properties of CO2, local mass and energy transfer between phases, and a two-phase sonic velocity model in the presence of finite-rate phase change. The model has been validated against the literature data for two-phase supersonic nozzles and overall ejector performance data. The results show that due to nonequilibrium effects and delayed phase change, the flow can choke well downstream of the minimum-area throat. In addition, Mach number profiles show that, although phase change is at a maximum near the boundaries, the flow first becomes supersonic in the interior of the flow where sound speed is lowest. Shock waves occurring within the nozzle can interact with the boundary layer flow and result in a ‘shock train’ and a sequence of subsonic and supersonic flow previously observed in single-phase nozzles. In cases with lower nozzle back pressure, the flow continues to accelerate through the nozzle and the exit pressure adjusts in a series of supersonic expansion waves.

Commentary by Dr. Valentin Fuster