Search

Effects of periodic disturbances on the development of a two-dimensional jet issuing from a contraction nozzle were studied experimentally. The disturbances were fed into the jet at the nozzle lips so as to excite the opposing shear layers with either symmetric or antisymmetric mode of velocity fluctuation. The forcing Strouhal number Ste, based on forcing frequency and nozzle width, was varied in the range of 0.08-1.2. The symmetric forcing at Stee0.4 and 0.8 increased the jet entrainment rate through frequency-halving processes. The maximal change of the jet width was achieved by the antisymmetric forcing at Stee0.1, which promoted the jet flapping motion in the region beyond the potential wedge.

Active control of jets emitted into a two-dimensional channel with insulated suction near the wall during monitoring with a neural network is described. The aim of this work is to simulate the possibility of controlling the flow pattern with a flip-flop phenomenon by suitable suction. In this case, the output of a neural network is estimated satisfactorily based on 3 flow patterns of teaching data. The periodic flow pattern at the Reynolds number Re =1.0x10 3 is reduced by a suction rate of less than 10% for primary jets. Then flow becomes stable as a result of feedback gain of the neural network. Since the control with suction need not be continuous, the power cost is reduced. The nonperiodic case at Re =1.0x 10 4 is difficult to control, but we achieved good control by with the use of a suitable suction rate with the neural network.

The effects of base suction on the near wake of a flat plate with a blunt trailing edge were examined by the discrete vortex method. The Strouhal number of vortex shedding was independent of the suction slit height in the range of suction flow rate below a certain critical value. Above the critical value, the Strouhal number progressively decreased with the suction flow rate, and the decreasing rate was dependent on the suction slit height. In this flow regime, two separating shear layers were alternately sucked into the slit. The Strouhal-number halving phenomenon which had been observed experimentally in the case of a circular cylinder was not found in the thick flat-plate wake.

A new method for estimating convection using a incompressible Navier-Stokes solver is presented. A least squares identification method was applied to obtain derivatives around a reference point. Higher order derivatives were also utilized in order to reduce artificial diffusivity caused by firstorder upwinding. An iterative method similar to SMAC (simplified marker and cell) was used for time marching, and least squares estimation was carried out to determine predictors of variables. Therefore, the estimation time can be neglected in the entire calculation. Flow around a cylinder and two-dimensional cavity flow were selected to examine the applicability of the current method. Static pressure distribution and velocity profiles agree fairly well with previously obtained data. However, it was found that artificial viscosity is not negligible, even if higher order derivatives were utilized.

The 5-equation model which represents weakly nonlinear two-dimensional thermal convection with various boundary conditions at the top and bottom walls has been constructed. Although the present system includes two control parameters of competing instabilities, the second parameter is fixed at unity, since the aim of this work is to obtain a model which is realistic compared with the Lorenz model but which maintains comparable simplicity. Examination of bifurcation structures of the present model obtained by increasing Rayleigh number as a parameter shows that there is a definite range of parameters in which the structures of orbits are strongly dependent on the boundary conditions such as slip/slip, slip/no-slip and no-slip/no-slip at respective top/bottom walls. In particular, a unique pattern for an inverse transitional region, which seems to be a "twin period-doubling bifurcation, " has been found for the first time in the case of no-slip/no-slip conditions.

A new method of unstructured mesh generation is presented. The advancing front method and Delauney triangulation have been commonly applied to it. In the former method, interior mesh nodes are generated based on the nodes on boundaries and parameters defined by the user. Unsuitable setting of parameters, however, may lead to failure in mesh generation. On the other hand, the latter method can accomplish mesh generation without failure, though the user has to set all the nodes. The present method, based on Delauney triangulation, requires the user to set only a few nodes on boundaries, and the other mesh nodes are generated automatically according to the curvature or proximity of boundaries. The perfomance of the presented method is illustrated in a region with multiple bodies.

Recently, many studies have been conducted on velocimetries by which flow field is measured by following tracer particles with CCD cameras, because noncontact measurements are possible at many positions. However, velocity does not correspond to particle speed unless particle diameters are sufficiently small, and measurement error is expected when the density of the tracer particle is different from that of fluid. Therefore, we propose a technique for estimating velocity using a Kalman filter for the case that tracers do not follow the flow. This estimation method is effective even if uncertainty of the particle diameter exists. It is shown that estimation of particle diameter is also possible. However, when fluid density changes, it becomes necessary to consider follow ability. This study focuses on the development of the method of estimating velocity by following several tracers simultaneously in flow fields. Characteristics of this technique have been numerically examined for a flow field that imitates a reciprocating internal combustion engine, and for a shock wave. We found that this new technique is effective in estimating flow velocities and density even if the tracers do not follow the flow.

In a thermal convection field constrained by right and left insulated walls, stable multiple solutions easily exist. In the case of a two-dimensional cavity with aspect ratio of 1.5, whose upper wall is cooled and lower wall is heated at a constant temperature, the two modes of thermal convection cell pattern are stable at Rayleigh number Ra of 2.0 × 105. That is, one-cell-mode and two-cell-mode. However, as Ra increases to 3.5 × 105, both modes become unstable, and appear in similar probability. The objective of this work is to demonstrate the possibility of controlling convection patterns under unstable conditions. We found that estimation using artificial neural network is effective to increase the probability of obtaining a desired mode, even in the slightly higher Ra of 2.5×105.

Chaotic mixing in two-dimensional square cavity flows with a time-periodic lid velocity was numerically simulated. To quantify the mixing in the cavity, the pseudo-Lyapunov exponent λ was measured. We selected the Reynolds number (Re) and amplitude of lid velocity (A) as a parameter which governs the flow field and mixing conditions, and investigated the relationship between these parameters and λ. The obtained results showed that neither Re nor A affects the value of period T which gives the maximum λ.

The Benard thermal convection field is known to have multiple solutions for a certain Rayleigh number condition. This means that several convection flow patterns can be stably sustained without an external control force. This study demonstrates that bimodal switching control is possible by means of flow state estimation. A neural network was used to recognize the transition state of the convection flow pattern. In order to change the flow pattern, an unstable feature of Benard convection was utilized. That is, the chaotic nature of wobbling around multiple solutions contributes to fast switching between stable modes. Numerical simulations were carried out to show the ability of the current control method.

A new method has been developed in order to estimate fluid velocity using tracer particles which do not follow the flow. The present PTV (particle tracking velocimetry) algorithm is an extension of Kalman filter-type velocimetry reported by Yagoh et al. (Yagoh, K., Ogawara, K. and Iida, S., Trans. Jpn. Soc. Mech. Eng., Vol. 57, No. 542, B (1991), p.3622) by which particle position measured in the presence of noise can be used to determine particle velocity. Although PTV is a common technique for flow measurements, one obvious shortcoming is noted. Namely, ordinary PTV algorithms could not be used when the particle density is different from the fluid density, because of ill-trackability of tracer particles. This may cause serious loss in accuracy when some forces cannot be neglected, for example, centrifugal forces of gravitational force. This study has proven that the new algorithm successfully identifies fluid velocity, and that it also has the ability to estimate particle diameter during the course of the tracking process.

An adaptive fuzzy reasoning strategy was studied to obtain fast convergence for a 2D incompressible Navier-Stokes solver. In this study, this reasoning was applied to control of the relaxation factor when the Poisson equation was solved by the point SOR method. Normalized residual and relaxation factors themselves were used for input parameters to the control map or lookup table. Several maps optimized for different flow conditions were prepared, and selected almost randomly. This map-selecting strategy made it possible for the flow solver to carry out fast and stable calculation. Two-dimensional thermal convection was used to examine the effectiveness of this scheme, and the solutions obtained with and without adaptive control were nearly equal.

A stochastic model which is used for simulating the time evolution of material line stretching in homogeneous turbulence is developed. Although the present model is similar to the Lagrangian random flight model, turbulent velocity is calculated for the Eulerian grid points considering space correlation in addition to time correlation. Scales of turbulence are estimated from turbulent energy (k) and dissipation rate (e). Deformation of material line in two-dimensional homogeneous turbulence is simulated by using this model and the result shows that spatial scale of turbulence affects on the distortion of material line.

Numerical simulations of a reacting 2-D plane shear layer are performed using the fast mixing model which we proposed in a prior paper. Shear flow is simulated by a Lagrangian particle tracing method with a stochastic model based on turbulent statistical quantities. The chemical process used here is "A+B→P", ignoring heat release. The validity of this model has been investigated for the following cases : 1) effect of reaction rate, 2) effect of lean-to-rich reactant ratio, 3) effect of "flip" case, in which chemical species of the high-speed side and the low-speed side are exchanged. This study proves the capability of the fast mixing model for simulating reacting shear flows.

An alternative method for particle tracking, which is needed in PIV (particle imaging velocimetry), has been developed by applying the Kalman filter. The method improves the precision in measurements of particle positions and of their velocity components. Using this method, it becomes possible to track each particle during more than hundreds of time steps. These effects and the propriety of the applied Kalman filter estimator are proved experimentally by calibrating the behavior of particles which are fixed on a rotating disk. Moreover, the method enables us to acquire data from individual tracer particles in a three-dimensional unsteady flow field.

A recent method "wavelet analysis" is applied as a tool for analyzing time sequential quantities derived from chaotic systems. As typical examples, the present method is applied to two well known systems, the Lorenz model and the Rossler model. The main procedure consists of calculating the local cross correlations between the signal to be analyzed and a predefined waveform called a "wavelet" at various time scales. This makes it possible to unfold the signal into both time and frequency, whereas in Fourier analysis, results are obtained in the frequency domain. The present paper mainly focuses on understanding the properties of this method, and shows a way to interpret the results.

A new algorithm for particle tracking has been developed in order to improve its trackability for flows in which considerable acceleration acts locally. This algorithm consists of two main stages. At the first stage, the particle tracking method using the Kalman filter "type P" is applied in order to estimate individual particle states ; then the measuring region is divided into subregions for the purpose of evaluating local statistics of the particle velocity and acceleration. At the second stage, a new type of Kalman filter which can take account of the local acceleration statistics is applied to tracking particles. Numerical simulations using a 2-D cavity flow are performed for accuracy tests of the present algorithm. As a result, it is proved that the present method achieves measurements with sufficient precision even when the flow direction changes rapidly due to the effect of large acceleration.

Conventional hot‐wire measurements have been made in the turbulent developing wake of a thin flat plate. The data obtained are primarily concerned with turbulence characteristics in the near‐wake region. Turbulence properties in the inner wake first increase up to a streamwise distance of 200–300 times the initial wall‐layer length scale, followed by a gradual decay farther downstream. The mechanism of this overshoot phenomenon is inferred from space‐time correlation data. Quasiperiodic motion similar to that in the far wake is observed as early as near the end of the near wake.

A new stochastic mixing model has been developed for chemically reacting turbulent flows. The Lagrangian particle tracing technique is used to represent turbulent diffusions. Concentrations of passive scalars are numerically estimated by the "fast mixing model", which proposes if two fluid particles approach each other within a certain prescribed distance as a result of stochastic processes, mixing occurs instantaneously and the concentrations become the average of the initial two values. The results obtained using this model were in good agreement with analytical solutions and gave an accurate diffusion coefficient.

Fuzzy control was applied to speed up the convergence of a Poisson equation solver for the numerical calculations of incompressible fluid. The optimizations of the control rule were carried out by means of a hybrid method of the conjugate gradient method and the improved random walk method. The residual of the equation and the relaxation parameter are employed as input for the controller. It is found that this method is effective in obtaining fast convergence. Moreover, it is clarified that this control does not affect the bifurcation characteristics of the solution. The changes in optimized relaxation parameter are similar to Bang-Bang control.

A new adaptive control method for fuzzy reasoning is proposed, in which the fuzzy controller works with a fluctuation generator for adaptation. The fluctuation generator which produces Gaussian noise is combined with antecedent parts of a fuzzy controller to fluctuate membership functions in each processing time. The learning parts of the system update the membership functions if the output from the controller is more suitable for faster convergence of the Poisson equation. Two-dimensional numerical simulation of the flow around a moving plate is conducted with the fuzzy controller using this new technique. The control variables, i. e., input to the controller, are the interval of the time step and the over-relaxation parameter of the Poisson equation. As a results it is found that this method is adaptive and time effective. The new adaptive method reduced the computational time to 80% compared that for a fuzzy controller without learning.

Two-dimensional numerical simulations of Taylor vortex flows considering acceleration rates of the inner cylinder were conducted to clarify the generating mechanisms of the primary, the secondary normal and the anomalous modes. We found that higher acceleration rates produce such strong vorticity for the cells on the end walls that a larger number of cells are induced toward the center. Slower increasing rates of the Re number promote diffusion of vorticity into the center of the cylinder. The modes are determined by the balance between induction and diffusion of the vorticity. The end cells degenerate in anomalous modes with certain acceleration rates. These weakened small cells are called "extra" or "hidden" vortices. We show that the extra vortices are end vortices which have become unstable and dissipate.

Steady and 2D-driven cavity flows were numerically simulated using the finite difference method. Pseudo-Liapunov exponents were obtained by measuring distances between particles initially located nearby. In the case where the moving lid was oscillated at a certain frequency, the particle paths showed a "folding and stretching" pattern and gave a positive exponent. We found that the pseudo-Liapunov exponent has its maximum at the oscillating period of 1.5T for Re=50.

The formation mechanism of streamwise vortices in the near field of the three-dimensional wall jet discharging from a circular nozzle along a flat plate is studied experimentally using a conditional sampling technique. Ensemble-averages of the lateral velocity component indicate the presence of large-scale horseshoe-like structures, whose legs are inclined and stretched to form the streamwise vortices in the mixing region of the jet. Based on the present result, a coherent structure model for the near field of the wall jet is proposed.

A Particle Tracking Velocimetry (PTV) algorithm based on Kalman filtering has been developed for the case where tracer particles may not follow the flow. This method is tested by numerical simulations to assure that the estimated velocities agree with the true fluid velocity. In order to confirm the velocity correcting ability of the extended Kalman filter, one-dimensional flows with significant density change were selected, that is, reciprocating flow in a cylinder with a compression ratio of 10.0 and stationary normal shock wave with known fluid density. For both cases, this new technique reproduces fluid velocities even though slip velocities between tracer particles and fluid are not negligible. To examine the effectiveness of this method in more realistic 2-D flow fields, we conducted numerical simulations of compressible flows in a channel with a small bump and a channel with a contracting section. The results show that 10-micron water particles in air flow can be used as tracers to obtain reasonable accuracy for velocity estimation. (Author)

An alternative particle tracking method for PIV (Particle Imaging Velocimetry) was developed by applying the Kaiman filter. Numerical simulations to track particles in a 2-D cavity flow were performed to investigate the effects of the present method. It was proved that the method can considerably improve the accuracy and the dynamic range of the PIV measurement.

A fast method to generate curvilinear coordinate systems with arbitrary line densities in the physical region is presented. There are no restrictions on the shape of the surrounding boundaries. Two numerical examples are given to show the validity of the method ; one is a rhomb as a typical shape of an inner boundary with many corners on its contour, and the other is a Karman-Trefftz airfoil as a typical case of a cusped body. In the latter case, in particular, an orthogonal curvilinear coordinate system can be obtained when a conformal mapping technique is used simultaneously. Moreover, it is expected that this grid generation procedure can be applied to a wide variety of other problems.

Coherent structures in a fully developed turbulent pipe flow are studied experimentally. The space-time correlations between wall pressure fluctuations and two components of velocity fluctuations, and behavior of those fluctuations conditionally sampled with respect to bursts are measured. According to the results, a factor of the averaged convection velocity of the burst to the center velocity is about 0.68, the bursts contribute to the production of higher frequency components of the wall pressure fluctuations, and further, there exist large scale coherent structures which extend from near the wall to the center region of the flow. Moreover, these results are compared with those of the fully developed turbulent plane boundary layer flow.

The longitudinal vortices in a two-dimensional stagnation region of a cylindrical blunt body are expressed as a perturbation with respect to the mean motion in terms of the amplitude and the reciprocal of the square root of Reynolds number, in order to theoretically analyze their stability up to the second order of magnitude. As a result, a specific point can be determined where the integration of the kinetic energy involved in the vortices becomes minimum and the first and second orders of the magnitude of disturbances would never diverge at a considerable distance from the wall, and further, the damping factor of -α takes the minimum value. Moreover, the wave number for α=0 and the minimum kinetic energy reasonably agree with observations.

A slightly non-linear case of stability is treated when the non-dimensional amplitude A0 of the disturbances in a two-dimensional stagnation region of a cylindrical blunt body is not so large that the derivative of A02 with respect to a non-dimensional time τ can be expressed as dA02/dτ = 2A02 (α0 + αlA02) where, α0 and αl denote the respective growth factors of the first and second degrees. The mutual combination between the signs of α0 and αl has an important significance for the problem of non-linear stability. In this paper, the results of αl < 0 is obtained by determining the value of αl such that the integration of the kinetic energy involved in the disturbances becomes minimum up to the higher order of magnitude. This result suggests that the flow in the stagnation region reaches the state of super-critical equilibrium when the Reynolds number exceeds the critical value.

A circular turbulent wall-jet along a convex wall was measured in comparison with that along a flat plate, and the flow characteristics of the three-dimensional wall-jet under a streamwise curvature effect was examined. By hot-wire anemometry, all velocity components were measured so that the components of turbulent intensity, the Reynolds stresses, and the mean streamwise vorticity were fiqured out. The main conclusions are as follows : 1) there are two pairs of secondary flows in the circular wall-jet along a flat plate, 2) an additional pair of secondary flows appears in the case of the flow along a convex wall 3) as an effect of curvature, the wall static pressure along the centerline decreases considerably, and 4) a significant increase of the vertical jet width compared with that of the spanwise one is observed.

In case of finite Reynolds number Re, the second order solutions of boundary-layer equation for the flow over a circular cylinder with separated wake are investigated in the vicinity of the stagnation point. The second order effects for wall shear stress Tw due to the curvature of the cylinder and the displacement of main flow are discussed by introducing the method for higher order boundary-layer solutions developed by Van Dyke. As a results, it is shown that Tw decreases proportionally to Re-1/2 and the effect of curvature amounts to 75 percent of the decreasing rate. Moreover, a new accurate method for solution of first order boundary-layer equation is proposed.

A study of balancing of flexible rotors with uniform cross sections supported by two bearings at both ends is performed by calculation, putting m as the number of critical speeds which are in the operating speed range. Using m+2 balancing planes, we can eliminate the static and dynamic unbalances at low speeds and the vibration components of the normal modes of the 1st, ……, mth order. When the rotors are simply supported, however, the vibration caused by some modal components of higher orders may appear in the high speed range. We can remove this vibration by employing m+4 balancing planes; we add the balancing operation to eliminate the two bearing forces at the highest speed. Regarding the rotors with spring supports and the rotors supported by springs and dampers, we can balance them by using balancing planes fewer than m+2 or m+4, when the supports are soft.

Since a theoretical treatment of the flow around a body with an arbitrary shape is usually difficult, only some of simple cases have been solved. In this paper, it is shown that a uniform flow around a body can be treated theoretically by an analysis of the combined flow of distributed three-dimensional doublets and uniform flow, so far as the body is of an axisymmetrical shape and does not have any sudden change on its surface. As an example, a clew of determining the position of static holes of a Pitot-static tube is also given by applying the analysis to the flow around a cylinder with hemispherical head. The analytical results obtained are compared with the combined flow of a three-dimensional source and uniform flow and also with an experimental result reported by Ower and Johansen.

A method to calculate the static deflection, the bending moment, the reactions at supports, etc. is proposed for continuous beams which have nonlinear supports. The calculation method is a kind of relaxation method. We linearize the system by replacing nonlinear supports with linear springs, and carry out the calculation of static deflection for the line-arized system. By iterating the linear calculations in which the linear spring constants are determined from the previous calculation result, we can obtain the solution which also satisfies the characteristics of nonlinear supports. The characteristics of the following supports are described as the examples of nonlinear ones: (1) a section of stern tube bearings made of lignumvitae, (2) a rigid support with backlash, and (3) a spring of which the end is separable from the beam when unloaded. As an example of numerical calculation, a calculation result for a propeller shaft of a ship is illustrated.

The near‐field measurement of fully turbulent jets discharged from a straight circular pipe indicates the presence of quasiperiodic velocity perturbations. The Strouhal number based on the predominant frequency falls in the range of the preferred mode for the jet with a potential core and has an asymptotic value of 0.38 in the high Reynolds number regime. Regular perturbations are not observed in the mixing region immediately downstream from the pipe exit.