The attachment of an inducer upstream of a main impeller is a powerful method to improve the suction performance. It is known, however, that the cavitation surge oscillation, which is focused in the present paper, occurs under the operation condition with partial flow rate and low suction pressure. The attempts to suppress the cavitation surge have been made by installing a ring-shaped obstacle plate just upstream of inducer. The ring-shaped inlet plates with various blockage rates of 14%, 27% and 39% have been tested to examine the effect of blockage rates on the suppression of cavitation surge for two kinds of inducers having two blades with the solidity of 2.0 and the different blade angles of 8° and 14° at the blade tip. In the present paper, the changes of the onset region of cavitation surge are clarified and the mechanism of cavitation surge suppression is discussed with the measured casing wall pressure distribution and inlet velocity distribution.
To elucidate the unsteady characteristics of a nonlinear pressure loss generated in a restricted flow, pressure drops across and flow rates through an orifice were precisely measured in sinusoidally oscillating oil flows. It has turned out that the unsteady relationship between the nonlinear pressure loss and the flow rate describes a closed loop turning around along the characteristic curve of the steady-state one in the counter-clockwise direction, which indicates that the change of the nonlinear pressure loss is delayed behind that of the flow rate in an unsteady orifice flow. The phenomenon occurs probably because the structure of turbulence in an orifice jet flow, where a nonlinear pressure loss is generated by energy dissipation, has inertia against a change of the flow rate and cannot follow it without a time lag. A mathematical model incorporating a constant time lag into the steady-state nonlinear pressure loss was proposed to simulate the unsteady characteristics of the nonlinear pressure loss, successfully explaining the long-term question in acoustics why the reactive part of an acoustic orifice impedance decreases as the amplitude increases.
Nowadays, rapid prototyping (RP) methods are widely used to produce wind tunnel testing models. Layer thickness is an important parameter that affects aerodynamic characteristics of wind tunnel models. This paper describes the effects of Layer thickness, using rapid prototyping, on aerodynamic coefficients to construct wind tunnel testing models. Three models were evaluated. These models were fabricated from ABSi by fused deposition method (FDM). The layer thickness was 0.178 mm, 0.254 mm and 0.33 mm. The surface roughness for each model was 25 μm, 63 μm and 160 μm (RZ) determined by PERTHOMETER2. A wing-body-tail configuration was chosen for the actual study. Testing covered the Mach no. range of Mach 0.3 to Mach 1.2 at an angle-of-attack range of -4° to +16° at zero sideslip. Coefficients of normal force, axial force, pitching moment, and lift over drag are shown at each of these mach numbers. Results from this study show that layer thickness does have effect on the aerodynamic characteristics; in general the difference between the data extracted from three models is less than 6 percent. The layer thickness does have more effect on the aerodynamic characteristics when mach number is decreased and has the most effect on the aerodynamic characteristics of axial force and its derivative coefficients.
Mean flow quantities have been investigated for the equilibrium boundary layer evolving over the rough wall with a roughness height that is proportional to streamwise distance. The wall shear stress measured by drag balance with a floating element device of a zero displacement mechanism. The local skin friction coefficient is independent of both streamwise distance and momentum thickness Reynolds number and has a value of 0.00826. The boundary layer thickness is proportional to the streamwise distance. The relative roughness height then maintains a constant value. The log-law profile has the same slope of that of the smooth wall turbulent boundary layer in the logarithmic liner part. The value of the roughness function increases in the streamwise direction. The wake parameter approaches to a constant value of approximately 0.70. In addition, the value is consistent with the result determined from the analytical method using the momentum integral equation and Coles's wake law. For the analytical result, the wake parameter decreases with increasing friction parameter.
An experimental study was conducted by performing pressure measurements and flow visualization to investigate unsteady flows inside a two-dimensional semi-open-type nozzle in a ship propulsion equipment directly driven by high-pressure gas. We found that the ejected gas phase and water-flow phase are separated clearly, and the interface between these phases behaves like waves. It was clarified by flow visualization with a high-speed motion camera and a circulating water channel that these interfacial waves change their shapes according to the water-flow velocity. The interfacial wavelength increases as a result of increasing water-flow velocity, and the mechanism that produces thrust on the nozzle wall changes. The thrust and flow patterns for intermittent gas ejection according to water-flow velocity were also clarified.
For revealing the transition process in a flat plate boundary layer subjected to a weak free stream turbulence, flow visualization and hot-wire measurements were performed. A weak free stream turbulence was generated by a turbulence grid mounted upstream of the contraction. The flow visualization clearly displayed a transition scenario in which a local two-dimensional wave packet rapidly forms a Λ shape structure and then breaks down to turbulence, resulting in the generation of a turbulent spot. Quantitative measurements performed by using a hot-wire anemometer also confirmed the existence of local Tollmien-Schlichting waves that agreed with the parallel linear theory in terms of their frequency, phase velocity, and the wall-normal distribution of band-pass-filtered fluctuations. For comparison, a boundary layer subjected to a moderate-intensity free stream turbulence was investigated. This investigation showed that streaky structures play an important role in the boundary layer transition, as shown by Matsubara et al. [J. Fluid Mech., 430, (2001), 149-168.] A drastic change occurred in the transition process and this change could be sensitively determined by employing the intensity and/or spectra of the free stream turbulence.
In the present study, to reveal the air entrainment mechanism into a suction pipe in a suction sump, the authors conduct flow-velocity measurements by UDM (Ultrasonic Doppler Method). Here, we consider the simplest geometry as a suction sump, that is, a straight channel with rectangle-cross section and a simple suction pipe near the end of the channel. Ultrasonic transducers are fixed outside the side, bottom and back walls with right/near-right angles and, we get three-dimensional time-mean velocity distributions and equi-vorticity contours. At first, measurement accuracy is checked by comparing velocity profiles by UDM with hydrogen bubble method. As a result, the authors show typical flow fields in the sump, and show the relation between flow pattern and air entrainment. Especially, we compare two cases where the air entrainment is often observed.
Three-dimensional turbulent flows past a square cylinder are simulated by the Finite Difference Lattice Boltzmann Method (FDLBM). We carry out the simulation in two approaches. First, we carry it out using fourth order numerical viscosity without any turbulent model, and second, we simulate using Dynamic Smagorinsky Model (DSM). These results are compared with those by the experiment conducted by Lyn et al. Numerical results by DSM agree with experimental ones and it is confirmed that this method is useful for numerical simulations of turbulent flows.
This paper presents the experimental result of the destruction of plankton using a cavitating jet generated by a two-dimensional nozzle. The destruction rate of plankton increased with the jet velocity and it reached almost 100% when the nozzle velocity was more than 30 m/s. However, the destruction rate of plankton eggs is lower than that of the plankton. The destruction is more effective when the cavitating jet impinges on the target body, as this causes cavitation bubbles to collapse intensely, and destroys plankton by high pressure generated at the collapsing stage.
In a plug transportation that is one of gas-solid two-phase flows, the prediction equations on particle velocity within a plug and pressure drop in a horizontal pipe have been formulated. The agreement between values calculated by these equations and experiments in which solid-air mass flow rate, pipe diameter, kinds of particle were changed was obtained. The difference between them is almost within 10%. In order to confirm the validity of supposition to derive these equations, particle velocity distribution within a plug in the directions of flow and radius, and particle velocity transformation from a stationary bed to a plug have been analyzed by high speed camera and PIV. The results show that there is no particle velocity distribution in a plug, and particles are accelerated uniformly in extra part of a plug. Namely, particles in a plug are fixed relative to each other and so they all move with the same velocity.
The propagation of pressure waves caused by a thermal shock in liquid mercury containing micro gas bubbles has been simulated numerically. In the present study, we clarify the influences of the introduced bubble size and void fraction on the absorption of thermal expansion of liquid mercury and attenuation of pressure waves. The mass, momentum and energy conservation equations for both bubbly mixture and gas inside each bubble are solved, in which the bubble dynamics is represented by the Keller equation. The results show that when the initial void fraction is larger than the rate of the thermal expansion of liquid mercury, the pressure rise caused by the thermal expansion decreases with decreasing the bubble radius, because of the increase of the natural frequency of bubbly mixture. On the other hand, as the bubble radius increases, the peak of pressure waves which propagate at the sound speed of mixture decreases gradually due to the dispersion effect of mixture. When the natural frequency of the mixture with large bubbles is lower than that of the thremal shock, the peak pressure at the wall increases because the pressure waves propagate through the mixture at the sound speed of liquid mercury. The comparison of the results with and without heat transfer through the gas liquid interface shows that the pressure waves are attenuated greatly by the thermal damping effect with the decrease of the void fraction which enhances the nonlinearity of bubble oscillation.
In concentrically rotating double cylinders consisting of a stationary outer cylinder and a rotating inner cylinder, Taylor vortex flow appears. Taylor vortex flow occurs in journal bearings, various fluid machineries, containers for chemical reaction, and other rotating components. Therefore, the analysis of the flow structure of Taylor vortex flow is highly effective for its control. The main parameters that determine the modes of Taylor vortex flow of a finite length are the aspect ratio Γ, Reynolds number Re. Γ is defined as the ratio of the cylinder length to the gap length between cylinders, and Re is determined on the basis of the angular speed of the inner cylinder. Γ was set to be 3.2, 4.8 and 6.8, and Re to be values in the range from 100 to 1000 at intervals of 100. Thus far, a large number of studies on Taylor vortex flow have been carried out; however, the effects of the differences in initial conditions have not yet been sufficiently clarified. In this study, we changed the initial flow field between the inner and outer cylinders in a numerical analysis, and examined the resulting changes in the mode formation and bifurcation processes. In this study, the initial speed distribution factor α was defined to be a function of the initial flow field and set to be 1.0, 0.999, 0.9 and 0.8 for the calculation. As a result, a difference was observed in the final mode depending on the difference in α for each Γ. From this finding, non-uniqueness, which is a major characteristic of Taylor vortex flow, was confirmed. However, no regularities regarding the difference in mode formation were found and the tendency of the mode formation process was not specified. Moreover, the processes of developing the vortex resulting in different final modes were monitored over time by visual observation. Similar flow behaviors were initially observed after the start of the calculation. Then, a bifurcation point, at which the flow changed to a mode depending on α, was observed, and finally the flow became steady. In addition, there was also a difference in the time taken for the flow to reach the steady state. These findings are based on only visual observation. Accordingly, a more detailed analysis at each lattice point and a comparison of physical quantities, such as kinetic energy and enstrophy, will be our future tasks.
A prototype system for feedback control of wall turbulence is developed, and its performance is evaluated in a physical experiment. Arrayed micro hot-film sensors with a spanwise spacing of 1 mm are employed for the measurement of streamwise shear stress fluctuations, while arrayed magnetic actuators of 2.4 mm in spanwise width are used to introduce control input through wall deformation. A digital signal processor with a time delay of 0.1 ms is employed to drive the actuators based on the sensor signals. The driving voltage of each actuator is determined with a linear combination of the wall shear stress fluctuations at three sensors located upstream of the actuator, and a noise-tolerant genetic algorithm is employed to optimize the control parameters. Feedback control experiments are conducted in a fully-developed turbulent air channel flow at the Reynolds number of Reτ=300. It is found that about 6% drag reduction has been achieved in a physical experiment for the first time. Through turbulent statistics measurements with LDV, it is also found that the Reynolds shear stress close to the wall is decreased by the present control scheme. A conditional average of a DNS database is also made to extract coherent structures associated with the present control input. It is shown that the wall-deformation actuators induce a wall-normal velocity away from the wall when the high-speed region is located above the actuator.
Direct numerical simulations of homogeneous isotropic turbulence laden with particles have been conducted to clarify the relationship between particle dispersion and coherent fine scale eddies in turbulence. Dispersion of 106 particles are analyzed for several particle Stokes numbers. The spatial distributions of particles depend on their Stokes number, and the Stokes number that causes preferential concentration of particles is closely related to the time scale of coherent fine scale eddies in turbulence. The influential Stokes number can be derived based on the Burgers' vortex model for coherent fine scale eddies. On the plane perpendicular to the rotating axis of fine scale eddy, number density of particle with Stokes number which is close to the influential Stokes number is low at the center of the fine scale eddy, and high in the regions with high dissipation rate of turbulent kinetic energy around the eddy. The maximum number density can be observed at about 1.5 to 2.0 times the eddy radius on the major axis of the fine scale eddy.
A numerical study is made to delineate the time-dependent spin-up flow of a stratified fluid in a cylinder. The time-dependent process of a rotating and stratified fluid, in response to the alteration of the rotation rate of the container, is studied. The change in rotation rate of the cylinder is described as an exponential function, with time constant tc. The classical instantaneous spin-up is modeled as a special case when tc is very small. Numerical solutions were acquired to the governing time-dependent, Boussinesq-fluid, Navier-Stokes equations. Detailed evolution patterns of flow are scrutinized to reveal both the angular velocity and meridional flows, as the value of tc covers a wide range. The numerically-obtained relative sizes of the dynamic effects are compared. The direct effect of fluid stratification on the global adjustment process is depicted. The role of viscous diffusion is discussed especially when the fluid stratification is substantial. In the case of a homogeneous fluid, the principal dynamic effects are confined to the boundary layers. However, for a stratified fluid, the viscous diffusion effect is not insignificant in much of the interior region.
Motion of small bubbles upstream of a grid spacer in a two by three rod bundle is experimentally investigated. Trajectories of single bubbles upstream of the grid spacer are recorded by using a high-speed video camera. An image processing method based on the combination of the Sobel filter and the Hough transform is developed to measure void distributions in dilute bubbly flows in the rod bundle. Mean liquid velocities upstream of the spacer are calculated based on 10,000 instantaneous velocity distributions measured by a Particle Image Velocimetry (PIV). Effects of the grid spacer on bubble motion are discussed based on the difference between the measured bubble motion and liquid velocity distribution. The main conclusions obtained are as follows: (1) just upstream of the grid spacer, bubbles are apt to migrate toward the rods and accumulate in the vicinity of the rod surface, and (2) lateral displacement of bubbles toward the rod wall is larger than that of streamline due to the centrifugal force induced by curved streamlines just upstream of the grid spacer.
A new experimental method is presented in which single small gas bubbles are generated in a liquid from a submerged orifice using pulsed ultrasound waves. Pulsed ultrasound waves having a frequency of 15 kHz and a maximum pressure amplitude of approximately 10 kPa are irradiated to a bubble growing from an orifice. Single air bubbles ranging from approximately 0.05 to 0.2 mm in radius are obtained in silicone oil (kinematic viscosity: 1 mm2/s) by using two orifices (0.02 and 0.04 mm in diameter) and by shifting the onset of the detachment-assistance pressure wave. The bubble deformation and detaching processes were visualized and analyzed using high-speed video imaging and direct numerical simulation. Consequently, it is revealed that the bubbles are forced to elongate upward due to the fast oscillatory flow of gas through the orifice, and the elongation causes the bubbles to detach from the orifice. The size of the bubbles at detachment is well estimated by employing a common spherical bubble formation model.
Viscosity has a great influence on nonlinearity especially in the transonic regime: however only preliminary research into the viscous effect on the aeroelstic response was conducted. Nonlinearity of the flowfield assumed to be the main source of flutter behaviours. This paper reports the results of intensive numerical simulations carried out on 2D aerofoil sections to investigate the effect of viscosity on flutter boundary and especially on limit cycle oscillation (LCO) characteristics. Unsteady Euler and Navier-Stokes flow solvers are developed for use in aeroelastic analysis with the moving grid method, while the aeroelastic response is calculated with the aeroelastic equations of motion in two degrees of freedom. The flow solvers are verified with simulations of pitching aerofoil. Two aeroelastic test cases ranging from subsonic to transonic flow regimes (M∞=0.3∼0.92, Rec = 12.3×106) were simulated in depth to obtain flutter boundaries as well as characteristics of LCO. The results indicate that the LCO occurs in the subsonic flow regime as well as in the transonic flow regime with far greater intensity caused by viscous effect. It has been shown that the weak divergence is found to be a transonic aeroelastic phenomenon.
Effects of an opposite secondary wall for intensification of the liquid jet that is formed through the near wall bubble collapse were numerically investigated. The axisymmetric form of the Navier-Stokes equations was solved to predict the flow conditions inside and outside the bubble. The volume of fluid approach was incorporated to capture the bubble surface. Putting an opposite secondary wall in the same distance from the bubble center as the primary wall distance, necking of the bubble was observed after the bubble reaches to its maximum volume. After the bubble splitting occurred, formation of two separating jets toward the primary and secondary walls was observed. It was notable that putting the secondary wall leads to the formation of the jet with a higher velocity magnitude. Also the jet velocity magnitude increased with increasing the secondary wall distance from the bubble center up to a certain limit. These analyses were performed for several initial inside to outside pressure ratios of the bubble. The results of the present simulations can be implemented to improve the tasks applying the near wall bubble collapse, like the collapsing bubble induced micro-pump.
The effects of the activation energy on the intrinsic instability of adiabatic and non-adiabatic premixed flames were studied by two-dimensional unsteady calculations of reactive flows based on the compressible Navier-Stokes equation. A sinusoidal disturbance was superimposed on a planar flame to obtain the relation between the growth rate and the wave number, i.e. the dispersion relation, and the burning velocity of a cellular flame generated by intrinsic instability. When the Lewis number Le = 1, the activation energy had no effects on the instability of adiabatic flames. In non-adiabatic flames, the growth rate and burning velocity decreased as the activation energy increased, because the reduction of temperature at the flame front had a great influence on the flame instability at large activation energies. When Le < 1, the activation energy had much effects on both adiabatic and non-adiabatic flames. As the activation energy increased, the growth rate and burning velocity increased drastically, because of the increase of the Zeldovich number. In addition, the unstable behavior of cellular-flame fronts was observed at large activation energies. When Le > 1, on the other hand, the growth rate and burning velocity decreased as the activation energy increased. This was because that the stabilizing influence of diffusive-thermal effects became larger. The obtained results showed that the activation energy played an important role in the intrinsic instability of adiabatic and non-adiabatic premixed flames.