Air/water two-phase flow performance of a centrifugal pump was investigated for five kinds of closed impellers, each of which has a different outlet or inlet blade angle. The results showed that (1) sudden pump head degradation due to gas accumulation in the impeller occurs at a lower air flow rate with increasing inlet or outlet blade angle of the impeller; (2) the pump head remains high even under the condition of gas accumulation when the impeller has large blade outlet angle. These results are qualitatively discussed including numerical calculations of one-dimensional two-phase flow.

View full abstract

Current progress in the field of supercomputers has made possible the direct numerical simulation of the gas-solid two-phase jet based on the Navier-Stokes equation and the Lagrangian equation of particle motion without the use of any assumptions or simplified models. In this study, three-dimensional Eulerian air velocities and Lagrangian particle trajectories are simultaneously calculated to describe the interaction between particles and air using a two-way method. Although the mesh size is roughly seven times the Kolmogorov microscale, the calculated turbulent characteristics of air and particles (mean velocity distributions and turbulent intensity distributions) are in fairly good agreement with experimental data obtained using laser Doppler anemometry. This means that the simulation well describes the motion of large-scale eddies which play an important role in the formation of turbulent gas-solid two-phase flow.

View full abstract

A numerical simulation is performed for a dispersed gas-solid flow in a vertical channel. Flow of the solid phase is obtained by calculating individual particle motions, i.e.by the Lagrangian method, while the gas flow is obtained by solving the equations of inviscid fluid. The Direct Simulation Monte Carlo (DSMC) method is used to take account of particle-to-particle collisions. Attention is paid to the case of low gas velocities and high solid loading, for which the solid phase strongly affects the gas flow field. Therefore the flow fields of gas and solid are determined simultaneously taking the interaction between both phases into consideration. It is found that the flow becomes unstable and inhomogeneous as the gas velocity decreases and the solid loading increases. Furthermore, effects of parameters such as the channel width and physical properties of particles on the instability and cluster formation are investigated.

View full abstract

Solid spherical shells of mm-order diameter can be applied in lightweight structural materials, buoyant catalytic agents, high-performance solid fuels and artificial organs. A device for sequential production of solid spherical shells using liquid-liquid-gas systems is comprised of a cylindrical container containing two immiscible liquids and a gas injection nozzle at the center of the bottom. By controlling the gas flow and the temperature field of the two liquid layers, liquid spherical shells are formed at the interface between the two immiscible liquids and they solidify during the upward motion. For stable production, it is necessary to correctly estimate the thermo-fluid flow and to control the temperature field in the two liquid layers with high accuracy. To develop a device for sequential production of solid spherical shells, natural convection heat transfer in the two immiscible liquid layers in the cylindrical container was studied. By solving the incompressible Navier-Stokes equations and the energy equations for the upper and lower liquids, the flow patterns and thermal field of the two liquid layers were quantitatively investigated. Suitable conditions for the sequential production of solid spherical shells were determined.

View full abstract

The dynamics of a laser-induced bubble in liquid nitrogen is studied experimentally. The motion of an almost spherical bubble, formed under appropriate control of optical conditions, is visualized by high-speed photography. Low subcooled liquid nitrogen at 78.0 K is used; the resulting high vapor pressure inside a bubble causes motion different from that in water at normal temperatures. Pressure is applied to liquid nitrogen to increase the degree of subcooling. An induced bubble grows similarly to a Rayleigh bubble under the effect of liquid inertia. The high vapor pressure, however, retards the collapse from the Rayleigh's solution, and makes the bubble surface rough by Taylor instability coupled with themodynamical effect. Estimation of the cavitation parameter in experiments of liquid nitrogen and water enables us to understand the transition from the motion being heat transfer dominant to that being liquid inertia dominant.

View full abstract

Propagation of shock waves in dilute bubbly liquids is investigated numerically. Governing equations for the bubbly liquid are formulated with emphasis on the radial and translational motions of the bubbles. The conservation equations for mass, momentum and energy of the bubble interior are solved directly in order to estimate precisely the effects of internal phenomena on the bubble motion. A numerical method, in which individual bubbles are tracked to estimate the effects of their volumetric changes and relative motions on the wave phenomena, is developed. For a wave process in a steady-state condition, simple relations, such as the propagation velocity, C_i, are derived. Numerical results under several conditions reveal that the terminal wave propagation velocity coincides with the propagation velocity in an isothermal equilibrium condition, C_i. The slippage between bubbles and liquid influences the time evolution of propagation velocity of the shock wave and the relaxation structure behind the wave. However, the slippage plays a minor role in the wave propagation Process.

View full abstract

Water vapor is produced in the singlet oxygen generator of a supersonic flow chemical oxygen-iodine laser as an undesirable by-product. Since the water vapor deactivates the excited iodine atoms very efficiently, it is removed using a water vapor trap before the singlet oxygen is mixed with the iodine. However, part of the water vapor passes through the trap, mixes with the iodine and expands through the supersonic nozzle. In the present study, a condensation model is proposed and the effect of the water vapor condensation due to the supersonic expansion is simulated numerically assuming that the mixing takes place instantaneously and the flow is one-dimensional. The condensation causes a reduction in water vapor concentration and, in this respect, the deactivation of the excited iodine atoms is suppressed. However, the latent heat released into the flow greatly suppresses the cooling in the supersonic expansion. As a result, the small signal gain coefficient is lowered considerably. However, the scattering and the absorption of the laser beam by the water droplets are negligibly small compared to the amplification.

View full abstract

Acoustic streaming is one of the nonlinear phenomena produced by strong sound waves. This type of streaming is driven by acoustic momentum flux in an attenuating sound field. A strong standing wave in a duct generated by finite-amplitude oscillation of the air column dissipating due to friction at the duct wall produces acoustic streaming. The velocity of streaming is estimated from the steady part of the second-order term of a perturbation expansion of a sinusoidal oscillation. However, finite-amplitude oscillation gives rise to shock-wave propagation in the duct. In order to estimate acoustic streaming produced by finite-amplitude oscillation, it is necessary to analyze the response of the oscillatory boundary layer to shock waves in detail. The present paper deals with numerical analysis of the acoustic streaming described above. The fourth-order spatial difference method is applied to two-dimensional analysis of acoustic streaming in this work. Calculated results show velocity distributions in the oscillatory boundary layer and structures of steady streaming for various amplitudes of oscillation.

View full abstract

Head-on collision of a planar shock wave with open-cell materials was studied experimentally. Three kinds of polyurethane foam are treated: foam 350×70×70, which is of low porosity (φ≒0.76) and high density (ρ_c=290kg/m³); foam 50×50×50, which is of high porosity (φ≒0.98) and low density (ρ_c=26kg/m³); and foam 13×13×13, which has the same density and porosity as foam 50×50×50, but has a different internal structure of foam material. Stress-strain curves of foams show high nonlinearity and hysteresis. The maximum stress value behind the foam just in front of the solid end wall, is larger than the reflected shock pressure at the normal solid wall. When a shock wave hits a foam surface, part of the shocked gas penetrates into the foam and interacts with the foam material. Measured stress histories at the foam base of the shock tube show stress which is significantly higher than that due to the pressure behind the reflected shock wave at the solid wall. In the high-density foam 350×70×70, the peak stress is the highest, the mobility of gas in the foam is very low, and its dynamics can be approximated by a single-phase problem. In the foam 50×50×50 and low-density foam 13×13×13, peak stresses are low and the peak value depends on the internal structure of the material. In these cases, the mobility of gas in the foam is high, and the dynamics must be treated as a two-phase problem.

View full abstract

In a previous paper (Bear et al. Fluid Dynamics Research, Vol.9, pp.155-164) a mathematical model was developed in order to analyze an abrupt pressure impact applied to a compressible fluid flowing through saturated, porous media under isothermal conditions. It was shown that during a certain time period following the onset of pressure change, the macroscopic fluid momentum balance equation conforms to a wave form. This paper presents a one-dimensional simple analytical solution to this wave equation. The wave equation is transformed to Euler's equation describing the motion of a "new" fluid with characteristics related to the former one.

View full abstract

Wave equations may be invoked when an abrupt change in pressure applied to a compressible fluid in saturated porous media⁽¹⁾ (Bear et al. Fluid Dynamics Research, Vol.9, pp.155-164). This paper presents a method leading to generalized fluid density, pressure and temperature. Using these generalized characteristics, we write Euler's equation as the one-dimensional expression for the analytical solution of equation of motion of the fluid. To obtain this, equation the porosity and the matrix strain and temperature of the solid are developed as explicit functions of pressure.

View full abstract

New inflow and outflow boundary conditions are proposed. These conditions are applied to two test problems. One is a problem in which a vortex convected by a uniform mean flow passes through an outflow boundary. The other is a two-dimensional spatially developing mixing layer. The new boundary conditions which give highly accurate results for both test problems consist of convective-viscous conditions and an approximate transport equation for pressure. The applicability of the convective-viscous conditions is verified by numerical tests which show that conventional boundary conditions do not approximate the velocity or the pressure fields on the boundary. The pressure transport equation proposed in this work is used to obtain the pressure on the inflow and outflow boundaries from a previous time step pressure field. As a result, we can conclude that the new boundary conditions are appropriate for direct numerical simulations of spatially developing flows.

View full abstract

A general procedure of solving incompressible viscous flow around multiple moving bodies was developed. An overlapped unsteady grid system which consists of a stationary main grid and subgrids attached to each moving body was employed. The incompressible N-S equations were calculated over this grid system. As a numerical scheme, the convective terms were approximated by the generalized QUICK method. A semi-implicit second-order two-step method was devised for unsteady time integration. The pressure was accurately solved by the newly developed Neumann Poisson solver. To make the procedure global, a mask function was introduced instead of dividing the main grid into several subregions. Triangular interpolation was applied for data communication between each grid. The accuracy of the procedure was verified by examining the flow around an oscillating cylinder. An application to flows around several moving bodies with multiple overlapped regions was made, which ensured the versatility of the present procedure.

View full abstract

Near-wall bursting phenomena have been detected both by the VITA technique and by a quadrant analysis. In addition to these detection schemes, visual burst detection has been performed in order to characterize the difference between VITA events and Q2 (second quadrant) events. It is shown that the spanwise yaw angle of visualized low-speed streaks for VITA events is greater than for Q2 events. Based on the experimental findings and the Euler equation, quasi-streamwise vortex models are proposed, and the results are compared with the experiments. Fluctuating velocity signals computed from the models are in good agreement with the experimental data for VITA events. Contour plots of the calculated axial velocity, spanwise vorticity and instantaneous Reynolds shear stress are presented. The generation processes of shear layer structures and Q2 and Q4 motions are demonstrated.

View full abstract

Universality of the Strouhal number in the wake flow of a cylinder over a wide range of Reynolds number, i.e., 5×10⁴<RC<10⁷, was considered for this paper. Eight cylinders with various surface roughnesses were used for the measurement. The universal Strouhal numbers considered were Roshko's number, Bearman's number, Griffin's number and what took the measured values of the width of wake streets for the reference length. These universal Strouhal numbers were calculated using the measured values of vortex frequencies, surface pressure distributions and drags. Comparisons were then made. It was made clear that Bearman's number was the most uniform of all the universal Strouhal numbers considered. It maintained a uniform value of 0.18 throughout the Reynolds number range. The reason for the nonuniformity of other Strouhal numbers was also considered.

View full abstract

We measured mean values of drags C_D, rms values of lifts C_L, and phase angles φ_L of lifts relative to body displacements on a rectangular cylinder. Cylinders with different side ratios of B/H=l, 2 and 3 were forced to oscillate transversely in a uniform flow at an amplitude of 14% of the shorter side H. The experiments were conducted in a water tunnel, with the forced oscillation Strouhal number St_C ranging from O.O to 0.5. Force measurements show that there exists a peak in C_D and C_L curves as St_C increases. The C_D values of the oscillating square cylinder after the peak are constant and much smaller than the values before the peak, while those of rectangular cylinders with B/H=2 and 3 are almost unchanged after the peak. The C_L values after the peak reach minimum values and then increase gradually due to flow inertia. The low-speed instability is aerodynamically featured by sharp decreases of C_D and C_L values after exceeding their peaks. Furthermore, flow visualization shows that the wake flow of an oscillating rectangular cylinder goes through three stages, i.e., full separation, alternate reattachment and full reattachment, with increasing St_C. Based on visualization, the flow features of vortex excitation and low-speed instability are discussed.

View full abstract

A separation and reattaching flow formed by the boundary layer separation at the edge of a blunt circular cylinder at Reynolds number of 1.0×10⁵ based on the main flow velocity U_∞ and the diameter of the cylinder D is studied. The time-mean and root-mean-square values of fluctuating surface pressure on the front surface and along the side face are presented. The reattachment length is 1.73D and is constant along the circumference. From the mean pressure distribution on the front surface, the thickness of the laminar boundary layer at separation is estimated to be 1.86×10^-4D. The low-frequency motion nL_r/U=0.011 at the leading edge, which is regarded as the flapping of the separated shear layer, was also found in the initial boundary layer upstream of separation. This pressure fluctuation inside the initial boundary layer, which is clearly identified without other disturbance generated by the shear layer rolling up, recirculating or feedback motion from the reattachment zone, is caused by upstream propagation of the fluctuating motion of the separated shear layer. In addition to this low-frequency flapping motion, the high-frequency velocity fluctuations of the shear layer next to separation, which demonstrate the Kelvin-Helmoltz instability and the evolution of the shear layer, are important for understanding and control of separation.

View full abstract

Simulation experiments in vitro related to flow through a vein and an airway of the lung were performed. The effect of longitudinal tension applied to thin-walled silicone rubber tubes on the relationship between the cross-sectional area of the tube and the transmural pressure, and on static and dynamic relationships between pressure drop and flowrate through the tube was investigated using five different kinds of specially designed tubes of the same size with different applied longitudinal tensions. As a result, the following facts were clarified: Applying longitudinal tension to the tube made it more difficult for the tube to be collapsed. The movement of the tube wall along the tube axis during the self-excited oscillation of flow was restricted and the most greatly collapsed portion shifted upstream. The amplitude of the oscillatory pressure drop was much larger than its time-averaged value. The time required for the tube to collapse gradually from a fully open state amounted to 80 to 85% of the period of the oscillation.

View full abstract

The pendelluft effect on gas transport and exchange in high-frequency ventilation was investigated using a simple computational model in which the pendelluft zone extending over lower airways consists of two branches in asynchronous motion. Calculations were carried out for the case in which the respiration is modulated by the pendelluft mode with amplitude ratio ranging from 0 to 2. The results show that increased regional tidal volume and sloshing of flow markedly improves the gas exchange rate, suggesting that gas exchange comparable to that in ordinary respiration is available in high-frequency ventilation.

View full abstract

There is a possibility that flow (thermal creep flow) induced by the temperature gradient along wall can occur in reduced pressure containers. Simulation of this type of flow is important in material technology, such as chemical vapor deposition or in the design of thermal devices that effectively utilize this phenomenon. The present study attempts to simulate this type of flow by the direct simulation Monte Carlo method. As a result, it is demonstrated that thermal creep flows with magnitudes of about 1O^-2 of the most probable molecular thermal speed are well simulated by using this method.

View full abstract

Confined jets of two-phase mist flow are important in the development of a two-phase ejector for the refrigeration cycle. However, the flow characteristics of the two-phase ejector have not been elucidated to date due to the nonequilibrium of velocity and temperature. In this study, the mixing characteristics of two-phase mist flow in the two-phase ejector were investigated experimentally. The pressure increases in the mixing section and the diffuser of the ejector were measured. The following results were obtained by comparison of the measured pressure increase with that calculated using a simple theory. Increasing the length and decreasing the diameter of the mixing section were effective for raising the pressure. The energy efficiency of the two-phase ejector used in this experiment was approximately 10%. This efficiency should be increased by improving the mixing characteristics of the ejector and the nozzle efficiency.

View full abstract

The gas velocity in the circulating fluidized beds of, for example, aluminum ore calciners, petroleum crackers and boilers is larger than the particles' terminal velocity. We propose the growing chain model, in which the particles form vertical chains in the fluidized bed, in order to determine theoretically the regimes of fast fluidization. Comparing the fluidization condition empirically recognized as the optimum operating condition for the circulating fluidized bed in boilers, the maximum fast fluidization gas velocity is found to coincide with the terminal velocity of an infinitely long chain, which is about twice as high as that of a single particle.

View full abstract

Hydrodynamically and thermally fully developed forced-convection heat transfer in cylindrical packed beds with constant wall heat fluxes is theoretically examined on the basis of our previous two-dimensional, distributed parameter model taking into account the effects of non-Darcy, variable porosity and radial thermal dispersion. Several system parameters including the Prandtl number, Pr, the ratio of the thermal conductivity of solid to that of fluid, κ, the bed radius to particle diameter ratio, Γ, and the particle Reynolds number, Re_p, are varied. The adequacies of some approximations to the governing equations are also examined in detail. It is found that the effects of the aforementioned system parameters on the relationship between the Nusselt number and Re_pPr are the same as those observed for packed beds with constant wall temperatures, but values of the Nusselt number are generally greater than those obtained under the boundary condition of constant wall temperature, as long as values of Γ, κ and Re_pPr are kept constant.

View full abstract

Heat transfer coefficients in nucleate pool boiling of binary mixtures were measured on a circular copper plate, 40 mm in diameter, for a wide range of heat fluxes and concentrations. Measurements were done for five mixtures of methanol/water, ethanol/water, methanol/ethanol, ethanol/n-butanol, and methanol/benzene, each in the saturated boiling under atmospheric pressure. Measured data were expressed in the form α=Cqⁿ for each concentration including pure components, where α is the heat transfer coefficient, q is the heat flux, and C and n are constants. The validity of available correlations for the boiling heat transfer of binary mixtures was assessed based on the measured data. Of the five correlations compared, the Stephan and Korner correlation is found the most appropriate although it needs an empirical constant that varies according to the mixture.

View full abstract

Premature dryout due to flow oscillation is a very important factor in designing boiling systems. The flow oscillation depends, in general, on system size and/or configuration, and therefore the relationship between the premature dryout and the flow oscillation has not been fully understood so far. In this investigation, a CHF experiment in a forced flow boiling channel under the oscillatory flow condition has been conducted. Numerical simulation has also been conducted based on the lumped-parameter model of the boiling channel. The simulation well represents the transient behavior of the dryout under the oscillatory flow condition.

View full abstract

A small-scale thermal diode for operation near room temperature was designed from a heat pipe with an asymmetrical structure, and its operational characteristics were investigated. Directional control of heat flow was achieved with a heat flow ratio of about 42 and an overall heat transfer coefficient of 1.1×10³ W/m²K even in the case of a temperature difference of less than 1 degree.

View full abstract

The effect of inert species on low-temperature piston-compression ignition is examined. Pseudo-air is prepared by replacing the inert species in air with pure argon. Lean mixtures are composed of leaded or unleaded gasoline as the fuel and the pseudo-air or real air as the oxidizer. Cool-flame ignition delay τ₁ does not differ between the case of using unleaded and that of using leaded gasoline. A difference between the effects of the two oxidizers can be found in the blue flame ignition delay τ₂ under considerably lean conditions at an equivalence ratio of 0.7. An anti-knock additive, tetramethyllead, does not seem to be responsible for this difference. At onset of cool flame and blue flame in a pancake chamber a sliced-pineapplelike structure which has a central core bulk and a surrounding ring portion near the cylinder wall appears. Only when air is used as the oxidizer, does a blue flame appear in the ring portion near the cylinder wall earlier than in the central core. Inert species such as nitrogen or argon might participate in the eIementary chemical reactions appearing in preflame periods. Argon is chemically inert, but nitrogen or nitrogen compounds probably participate in the preflame reactions in the case of low-temperature ignition.

View full abstract

The transition from laminar to turbulent mode of an ethylene jet flame was investigated using the two-dimensional instantaneous photography of turbulent eddies by a laser-light sheeting method. Observations were made of the eddies around the break point in the fuel flow and the flame. The results show that in the laminar region the fuel flow is curved due to instability in the shear layer, whereas the outer soot layer has little curvature because of the high viscosity in the hot layer. In the transient region, eddies generated in the fuel flow deform the outer soot layer. Numerical calculations were performed to predict fluid motions due to interaction between density and pressure gradients in the flame boundary. The results show that the pressure gradient in a medium of varying density generates the vorticity along the flame. Deformation and stretching of the flame boundary take place once the vorticity becomes stronger than the dissipation due to viscosity.

View full abstract

An asymptotic analysis is applied for the extinction of diffusion flames in supersonic stagnation-point flow. (It can be considered as the Tsuji-burner in a supersonic flow.) If airflow velocity is subsonic, the flame temperature decreases as airflow velocity increases because the flame stretch rate increases, and finally flame extinction takes place at a certain airflow velocity. However, in the case of supersonic airflow, the region in which a diffusion flame can be established between a detached shock wave and a porous body reappears, and flame extinction occurs when the airflow velocity decreases. This indicates shock heating has a stronger effect than high stretch-rate flow and dominant parameter for the phenomena is a temperature behind a detached shock wave. Moreover, it is clarified that the flame temperature rises sharply with Mach number of airflow.

View full abstract