Search

In recent years, to predict the performance and the aerodynamic sound of box fans with high accuracy, numerical analysis has been necessary for the transitional boundary layer on the blades. In this study, two types of box fans with a baseline fan and a high-load fan, where the boundary layer that develops on the suction surface of the rotor blade transitions to turbulent flow, were targeted for large eddy simulation (LES) analysis and acoustic analysis using different grid resolutions to predict the far-field sound. The box fan in this study has an impeller of 180 mm outer diameter and five blades, and the Reynolds number based on the diameter and the tip velocity is 3.38 × 105. The decoupled method with LES for incompressible flow and computational aeroacoustics (CAA) based on a dipole sound source was performed. The computational grid resolutions had 25 million grid points (Case 1), 52–56 million grid points (Case 2 and Case 4), and 420–450 million grid points (Case 3 and Case 5). The performance of the fans and the sound pressure level (SPL) were compared with each case and experiment. The performance was in good agreement with the experiment in higher grid resolution cases. Boundary layer transition on the blade surface was predicted as the main source of aerodynamic sound generated from the box fan. Increasing the grid resolution improved the prediction accuracy of the SPL in the frequency range corresponding to the scale of the turbulent eddies captured by the grid resolution.

Aerodynamic noise in high frequency band related in the flows around A-pillar is easily propagated to the cabin of the automobile. Understanding of the characteristics of the aerodynamic noise is needed to reduce the cabin noise. Moreover, understanding of the separated shear layer which is strongly affected by the condition of the step hight H to the boundary layer thickness δ is needed. In this study, far field sound and correlation measurement of flow velocity are measured to investigate the characteristics of aerodynamic noise and the relationship between H/δ and turbulence length scale L as eddy scale for the forward facing step. As a result, multiple gradual peaks in more than 1 kHz are observed in the aerodynamic sound spectra. Furthermore, it is found that turbulence length scale L increases for increasing of H/δ and turbulence length scale to step height for H/δ = 2.0 is almost equal to that for H/δ = 3.0. Then, it is clear that spectral peaks are strongly affected by step height. And it is confirmed that H/δ = 2.0 and 3.0 are equal in turbulence length scale to the step height.

To clarify the mechanism for aerodynamic sound to be generated from a lifting surface placed in a flow with turbulence, wind-tunnel experiments and numerical simulations have been carried out for a flow around an airfoil subjected to the wake of a circular cylinder. The test airfoil has the NACA0012 profile with a chord length of 150 mm and a spanwise length of 500 mm, and it is set in a flow with the cylinder wake at angles of attack of 0 degree and 9 degrees. The wind velocity is set to 30 m/s, which results in an airfoil Reynolds number of 3.0×105. The circular cylinder is set 100 mm upstream of the leading edge of the airfoil, and its diameter as well as installation position in the transverse direction are varied. The numerical simulations are composed of large eddy simulation (LES) and aeroacoustical simulations that solves the Lighthill equation in the frequency domain with the Lighthill tensor computed by the above-mentioned LES as the acoustical source terms. The sound pressure level is also computed by the Curle’s equation with the fluctuation in the airfoil lift force computed by the LES as the acoustical source and compared with the experiments. The computed static-pressure distributions on the airfoil surfaces and aerodynamic sound spectra are in good agreement with the measured equivalents. The pressure fluctuation near the leading edge of the airfoil is remarkably high when the circular cylinder is placed upstream. When they flow near the leading edge of the airfoil, the vortices in the cylinder wake are stretched due to the acceleration of the main flow, form strong source of sound, and radiate intense sound in the upstream direction.

Sound generated aerodynamically from a flow around an airfoil subjected to inflow turbulence is investigated experimentally as well as numerically to identify the dominant source of the sound. The test airfoil has NACA0012 wing section with a chord length of 150 mm and a spanwise length of 500 mm. The wind speed is set to 30 m/s, which results in an airfoil Reynolds number of 3.0×105. Wind tunnel experiments are conducted with Active Turbulence Generator (ATG) set at the nozzle exit to control the turbulence intensity and the eddy scale of the inflow turbulence, independently. The former is varied from 10 % to 20 % while the latter is varied from 300 mm to 1600 mm. Large eddy simulation (LES) with turbulent grids is also performed to investigate the flow field around the airfoil in further detail, for which the turbulence intensity is 10 % and the eddy scale is 3.75 mm. As a result, we have found that the pressure fluctuation near the leading edge of the airfoil is remarkably high, and therefore, it is most likely the cause of the considerable increase in the sound generated from the airfoil. In addition, the correlation between the airfoil lift force and the angle of attack for the airfoil subjected to the inflow turbulence is not very high, and therefore, the change in the lift force due to the change in the angle of attack is least likely to the cause of increase in the sound.

The flow structures of a submerged vortex that appears in a model pump sump have been fully clarified by performing large eddy simulation (LES) of a model vortex in a simplified computational model. The computational model had a sufficiently fine grid that could resolve the vortex core. The model sump is composed of a 2,500 mm-long water channel of rectangular cross section with a width of 300 mm and a water height of 150 mm and a vertical suction pipe with a 100 mm diameter installed at its downstream end. Our previous large eddy simulations (LES), which used approximately 2 billion grids and were applied to the whole pump sump, has fully clarified the origin and formation mechanism of a submerged vortex and an air-entrained vortex. In these computations, however, the static pressure in the vortex core decreased by only 5 kPa at a channel velocity of 0.37 m/s. The decrease in the static pressure was far smaller than the one for which one can expect initiation of cavitation in the vortex core. In the corresponding experiment, however, appearance of a submerged vortex was confirmed by the occurrence of cavitation in the vortex core. Therefore, the decrease in the static pressure is most likely to be underpredicted in our previous LES. Insufficient grid resolution was assumed to be one of the reasons for this underprediction. In the present study, LES with a sufficiently fine grid was applied to a simplified computational model that represents the stretch of a submerged vortex under a constant acceleration of the vertical velocity. Vertical and tangential velocities obtained by averaging those profiles of a submerged vortex computed in the previous LES were prescribed at the bottom wall of the computational domain as the inflow boundary conditions. In the present LES, the static pressure has decreased by more than 100 kPa. In addition, parametric studies with different initial swirl numbers varied from 0.08 to 10.9 have fully clarified the behavior of a submerged vortex. It is found that a strong submerged vortex appears only at a relatively small range of the swirl-number from 0.8 to 2.

The origin and formation mechanism of a submerged vortex and an air-entraining vortex have been fully clarified by large-eddy simulation (LES) that used approximately 2 billion hexahedral elements with maximum resolution of 0.255 mm and was applied to the internal flows of a model pump sump. The model pump sump is composed of a 2,500 mm-long water channel of rectangular cross section with a width of 300 mm and a water depth of 100 mm and a vertical suction pipe with a 100 mm diameter installed at its downstream end with an offset of 10 mm from the centerline of the rectangular channel. At the upstream end of the channel, a uniform velocity of 0.37 m/s is given. LES with different wall boundary conditions have revealed that the origin of a submerged vortex is the mean shear of the approaching boundary layers that develop on the bottom and side walls of the pump sump. From detailed investigations of LES computed for a long time period of 16 seconds have revealed that deviation of the mean flow that approaches the suction pipe triggers conversion of the axis of the vorticity that was originally aligned to the lateral direction in the approaching boundary layers to that aligned to the vertical direction. The local acceleration of the vertical flow stretches the afore-mentioned vertical vortex, which results in formation of a submerged vortex. The separated flows downstream of the suction pipe generate vertical vorticity, and forms an air-entraining vortex when such a vortex is sucked into the suction pipe. Computations with a different bellmouth height and a different water-surface height have supported the above mentioned origin and formation mechanism of these vortices.

For designing a high-performance mixed-flow pump, control and suppression of performance-curve instability is crucial. For doing so, clarification of the essential flow physics that leads to the performance-curve instability is crucial in addition to an accurate prediction of the performance curve. We investigated internal flows of a mixed-flow pump, with performance-curve instability at round 60% of the designed flow rate, by using Large Eddy Simulation (LES). In the previous report, we clarified that drop of the Euler's head caused by stall near the tip of the impeller's blades is the major cause of the performance-curve instability. In the present report, we investigated in detail unsteady internal flows of individual blade passage near the head-drop point. As a result, we clarified the followings. (1) When the flow rate through a blade passage is low, the Euler's head is high for that particular passage, and this condition propagates in the direction of the impeller rotation for the whole impeller while the impeller makes approximately 20 revolutions. (2) Upstream flow of the stalled blade passage is deviated to the downward neighboring blade passage, resulting in increase in the load of the neighboring blade near its trailing edge. The neighboring blade passage will then be stalled. (3) When the flow rate in a stalled blade passage reaches the minimum, the passage starts to recover from stall while the pressure gradient in the streamwise direction will decrease.

A technique for high-accuracy predicting performance curve instability in pumps and clarification of the phenomenon in order to control the instability is required for designing high performance pumps. We analyzed internal flow of a test mixed-flow pump by using Large Eddy Simulation (LES) and clarified that Euler's head drop caused by stall near the tip of impeller blade is a dominant factor of this test pump's instability. We looked carefully at internal flow of impeller before and after the head drop, and also in the middle of the head drop. When the Euler's head drops, recirculation on the leading edge of the tip of impeller blade extends towards trailing edge. Angular momentum between tip and mid span decreases, and angular momentum near the tip is high. Work drop of the impeller on the side of leading edge caused by the stall on leading edge exceeds angular momentum increment caused by decreasing in flow rate on trailing edge, and so the angular momentum on the trailing edge drops. At that situation, the impeller blade load on the tip side shows state of stall. And, the stall fields occurred on the trailing edge cover whole impeller blade passages near the tip. When the head is in the middle of drop, only some impeller blade passages are covered with the stall fields.

To investigate the mechanism of noise generation by a train-car gap, which is one of a major source of noise in Shinkansen trains, experiments were carried out in a wind tunnel using a 1/5-scale model train. We measured velocity profiles of the boundary layer that approaches the gap and confirmed that the boundary layer is turbulent. We also measured the power spectrum of noise and surface pressure fluctuations around the train-car gap. Peak noise and broadband noise were observed. It is found that strong peak noise is generated when the vortex shedding frequency corresponds to the acoustic resonance frequency determined by the geometrical shape of the gap, and that broadband noise is generated at the downstream edge of the gap where vortexes collide. It is estimated that the convection velocity of the vortices in the gap is approximately 45% of the uniform flow velocity.

Large-eddy simulation of compressible transitional flows in a low-pressure turbine cascade is performed by a 6th-order compact finite-difference and 10th-order filtering method. Numerical results without free-stream turbulence and those with about 5% free-stream turbulence are compared. Based on the computed results, detailed investigations are presented on the effects of the free-stream turbulence on the boundary layer transition, and on the behavior of the pressure waves that originate near the trailing-edge along with the effects on the separation/transition of the boundary layer. Also, based on Snapshot Proper Orthogonal Decomposition (POD) Analysis, dominant behaviors of the transitional boundary layers are investigated.

The present paper treats the backflow vortex structure observed at the turbomachinery inlet at reduced flow rate. It is caused by the roll-up of the shear layer between swirling backflow and axial main flow. In order to verify this, a simple model test was carried out in which the effect of impeller was represented by an axisymmetric swirling backflow. In the present paper, the flow field of a simplified model test is simulated by using LES calculations to investigate detailed flow structure. The computed results are compared with experimental results.

This article describes the large-eddy simulation (LES) of the internal flows of a high–specific-speed, mixed-flow pump at low flow-rate ratios over which measured head-flow characteristics exhibit weak instability. In order to deal with a moving boundary interface in the flow field, a form of the finite-element method in which overset grids are applied from multiple dynamic frames of reference has been developed. The method is implemented as a parallel program by applying a domain-decomposition programming model.