Effects of Viscosity, Particle Size, and Particle Shape on Erosion in Gas and Liquid Flows

Although solid particle erosion has been examined extensively in the literature for dry gas and vacuum conditions, several parameters affecting solid particle erosion in liquids are not fully understood and need additional investigation. In this investigation, erosion ratios of two materials have been measured in gas and also in liquids with various liquid viscosities and abrasive particle sizes and shapes. Solid particle erosion ratios for aluminum 6061-T6 and 316 stainless steel have been measured for a direct impingement flow condition using a submerged jet geometry, with liquid viscosities of 1, 10, 25, and 50 cP. Sharp and rounded sand particles with average sizes of 20, 150, and 300 μm, as well as spherical glass beads with average sizes of 50, 150 and 350 μm, were used as abrasives. To make comparisons of erosion in gas and liquids, erosion ratios of the same materials in air were measured for sands and glass beads with the particle sizes of 150 and 300 μm. Based on these erosion measurements in gas and liquids, several important observations were made: (1) Particle size did not affect the erosion magnitude for gas while it did for viscous liquids. (2) Although aluminum and stainless steel have significant differences in hardness and material characteristics, the mass losses of these materials were nearly the same for the same mass of impacting particles in both liquid and gas. (3) The most important observation from these erosion tests is that the shape of the particles did not significantly affect the trend of erosion results as liquid viscosity varied. This has an important implication on particle trajectory modeling where it is generally assumed that particles are spherical in shape. Additionally, the particle velocities measured with the Laser Doppler Velocimetry (LDV) near the wall were incorporated into the erosion equations to predict the erosion ratio in liquid for each test condition. The calculated erosion ratios are compared to the measured erosion ratios for the liquid case. The calculated results agree with the trend of the experimental data.

Simulation of Liquid Water Evaporation in GDL for PEMFC Under Gas Purge Condition

In automotive Polymer Electrolyte Fuel Cell (PEFC) system, dry gas purge operation is needed at shutdown condition in order to remove the liquid water in gas diffusion layer (GDL) and to reduce the oxygen diffusion inhibition by liquid water in GDL. However, exceed drying operation leads to degradation of electrolyte membrane because of little water content. Therefore, drying process has to be optimized. In this study, various GDL structure with unique fiber orientation were simulated by numerical analysis, and the real GDL structure was reconstructed by X-ray CT image of carbon paper GDL. Next, our past two-phase network model was improved to include phase change effect. The multi-block two-phase network model based on an actual structure was developed by a direct 3D networking porous structure. As results, the evaporation interface area depended on the porous structure of GDL, and the overall evaporation rate of homogeneous GDL which has uniform structure was 1.5 time higher than that of heterogeneous GDL because of the difference of this interface area. In addition, in the case of rib and channel, liquid water under channel evaporated faster than that under Rib. It is very important to control the drying operation in order to prevent the excess membrane drying.

Motion of Descending Solid Particles and Local Flow Around the Particles in Downward Turbulent Water Duct Flow (Keynote)

The Stokes number, the ratio of the particle time scale to flow time scale, is a promising quantity for estimating changes in statistics of turbulence due to particles. First, we explored the Stokes numbers in some recent studies. Secondly, we discussed the results of our direct numerical simulation for turbulent flow with a high-density particle in a vertical duct. In the discussion, we defined the particle Reynolds number from the mean fluid velocity in the near-particle region at any time. We evaluated a new local Stokes number for the particle. It is found that the Stokes number is effective for the prediction of the distance between the particle center and one wall. Finally, we carried out experiments for turbulent water flow with aluminum balls of 1 mm in diameter in a vertical channel. The motions of aluminum balls and tracer particles in the flow were captured with a high-speed video camera. We found that the experimental results for the time changes in the wall-normal distance of the ball and the particle Reynolds number for the ball are similar to the predicted results.

On the RANS Modeling of Turbulent Airflow Over a Simplified Car Model

The need for reliable CFD simulation tools is a key factor for today’s automotive industry, especially for what concerns aerodynamic design driven by critical factors such as the engine cooling system optimization and the reduction of drag forces, both limited by continuously changing stylistic constraints. The Ahmed body [1] is a simplified car model nowadays largely accepted as a test-case prototype of a modern passenger car because in its aerodynamic behavior is possible to recognize many of the typical features of a light duty vehicle. Several previous works have pointed out that the flow region which presents the major contribution to the overall aerodynamic drag, and which presents severe problems to numerical predictions and experimental studies as well, is the wake flow behind the vehicle model. In particular, a more exact simulation of the wake and separation process seems to be essential for the accuracy of drag predictions. In this paper a numerical investigation of flow around the Ahmed body, performed with the open-source CFD toolbox OpenFOAM®, is presented. Two different slant rear angle configurations have been considered and several RANS turbulence models, as well as different wall treatments, have been implemented on a hybrid unstructured computational grid. Pressure drag predictions and other flow features, especially in terms of flow structures and velocity field in the wake region, have been critically compared with the experimental data available in the literature and with some prior RANS-based numerical studies.

Large Eddy Simulation on the Unsteady Aerodynamics of a Heavy Duty Truck in Wind Gust

Unsteady aerodynamic forces acting on a full-scale heavy duty truck were investigated using a large-eddy simulation technique. The numerical method adopted was first validated on a static condition measured at the DNW German-Dutch wind tunnels. After the correction of the blockage ratio in the wind tunnel, the drag coefficient obtained by our numerical method showed good agreement with the experimental data within the errors of less than 5%. Effect of an air deflector mounted on the top of a cabin was also discussed. Then the method was applied to non-stationary conditions in which the truck was subjected to ambient perturbation of approaching flow. The perturbation of the flow is a model of atmospheric turbulence and sinusoidal crosswind velocity profiles were imposed on the uniform incoming flow with its wavelength comparable to the vehicle length. As a result, it was confirmed that when the wavelength of the crosswind is close to the vehicle length, averaged drag increases by more than 10% and down-force decreases by about 60%, compared with the case without perturbation.

The Influence of Major Diameter Solid Particle on the Double-Channel Pump Performance

Based on mixture model, the numerical simulation of solid-liquid two-phase flow in a double channel pump (Specific speed ns = 81) was carried out. The effects of particle diameter, particle volume fraction and flow rate on solid volume concentration distribution, relative velocity distribution and abrasion characteristics were studied. The results reveal that in the impeller, more particles concentrate at the nut of the shaft end and the edge of the impeller outlet. So those regions are worn seriously. The abrasive types are sliding wear on the impeller outlet edge and impact wear on the nut respectively. In the wall of the volute, the concentrated areas of particles move round the anticlockwise direction when the mixture flow rate is larger. The reason is the mixture velocity is larger as the flow rate increases, and meanwhile the centrifugal force and gravity force are invariable. So the particles move round the impeller rotational direction consequently. In the volute, particles concentrate on the tongue and wall region, especially on the sections I, II, V and VII. So the areas are easily worn out. The abrasive type is the heavy sliding wear in the volute wall. Numerical simulation results are consistent with the actual situation. It follows that the calculating method is feasible.

Analysis of Turbulence Structure and Void Fraction Distribution in Gas-Liquid Two-Phase Flow Under Bubbly and Slug Flow Regime

Accurate analyses of turbulence structure and void fraction distribution are quite important in designing and safety evaluation of various industrial equipments using gas-liquid two-phase flow such as nuclear reactor, etc. Using turbulence model of two-phase flow and models of bubble behaviors in bubble flow and slug flow, systematic analyses of distributions of void fraction, averaged velocity and turbulent velocity were carried out and compared with experimental data. In bubbly flow, diffusion of bubble and lift force are dominant in determining void fraction distribution. On the other hand, in slug flow, large scale turbulence eddies which convey bubbles into the center of flow passage are important in determining void fraction distribution. In turbulence model, one equation turbulence model is used with turbulence generation and turbulence dissipation due to bubbles. Mixing length due to bubble is also modeled. Using these bubble behavior models and turbulence models, systematic predictions were carried out for void distributions and turbulence distributions for wide range of flow conditions of two phase flow including bubbly and slug flow. The results of predictions were compared with experimental data in round straight tube with successful agreement. In particular, concave void distributions in bubbly flow and convex distribution in slug flow were well predicted based on the present model.

Diagonal Flow Pump Impeller With NACA65 Series Blade (Correction of Blade Geometry)

Using the design method based on the design for axial-flow type turbomachine, the diagonal flow pump impellers were designed for two cases of the centrifugal effect parameter a. In addition, three-dimensional Navier-Stokes numerical calculations for single-phase were conducted in order to examine the tendency of the suction performance as well as the head performance. The head increases from the NS calculation of the impeller are the same between for a = 0.4 and for a = 1.0 because major specification are the same between for a = 0.4 and for a = 1.0. For the minimum pressure on the rotor blade, however, there is a difference between for a = 0.4 and for a = 1.0. The value of minimum pressure for a = 0.4 is −324 kPa, whereas the value for a = 1.0 is −294 kPa. The blade geometry for a = 1.0 is better than the one for a = 0.4 in terms of the suction performance because the trough of the minimum pressure is shallower for a = 1.0 than a = 0.4. Furthermore, Navier-Stokes numerical calculations were also conducted for off-design flow rate. For all cases in this paper, the minimum pressure on the rotor blade occurred at both near the leading edge and near the tip on the suction side of the blade. In addition, for all cases in this paper, the blade geometry for a = 1.0 is better than the one for a = 0.4 in terms of the suction performance.

A Dynamic Procedure for Wall-Modeled Large-Eddy Simulation of High Reynolds Number Flows

This paper addresses the error in large-eddy simulation with wall-modeling (i.e., when the wall shear stress is modeled and the viscous near-wall layer is not resolved): the error in estimating the wall shear stress from a given outer-layer velocity field using auxiliary near-wall RANS equations where convection is not neglected. By considering the behavior of turbulence length scales near a wall, the cause of the errors is diagnosed and solutions that remove the errors are proposed based solidly on physical reasoning. The resulting method is shown to accurately predict equilibrium boundary layers at very high Reynolds number, with both realistic instantaneous fields (without overly elongated unphysical near-wall structures) and accurate statistics (both skin friction and turbulence quantities).