An improved CFD simulation for investigation of the sand particles flow behavior in the sand shooting process

Highlights

• The EMMS-based drag model is introduced in sand shooting simulation.

• Validation of simulation by high-speed camera recording experiments is illustrated.

• Specularity coefficient affects sand particles flow behaviors.

• EMMS model get a well-prediction for smaller particles and dense flow.

Abstract

The flow behavior of sand particles has been examined by computational fluid dynamics (CFD) and experimental validations in sand shooting process. To avoid over-estimating the drag force between gas and solid phases, the energy minimization multiscale model(EMMS)has been incorporated. An applicable value of grid size is determined by grid dependency study. Three specularity coefficients are simulated to get an insight into the influence on the flow hydrodynamics of sand particles. The specularity coefficient affects tangential velocity and volume distribution of sand particles. No slip wall boundary condition is in better agreement with experimental data. With validation of sand shooting experiment, agreement is achieved between experiment images and simulation results. A comparison between EMMS drag model and Gidaspow model has been discussed in terms of simulating particles flow behavior with different particle diameters. The EMMS drag model has an accuracy prediction especially for smaller particles and dense flow.

Introduction

The sand shooting process is the most widely used technique in metal casting industry for manufacture of sand cores and has a significant affected on qualities of sand cores [1]. Due to the complexity of sand shooting process, the mechanisms of sand particles flow behavior and the interaction between air and sand particles are not fully understood. Therefore, a further investigation of the instantaneous gas-particle flow behavior is still necessary in the sand shooting process [2].

Nowadays, computational fluid dynamics (CFD) has become a powerful tool to investigate the hydrodynamic behavior of gas-solid flow [3,4]. CFD models for gas-solid flow can be summarized in three models at different scales [5]: direct numerical simulation(DNS) [[6], [7], [8], [9], [10]], discrete particle model(DPM) [[11], [12], [13]],and two-fluid model(TFM). The TFM which is based on Eulerian-Eulerian theory, treats both particles and gas phase as interpenetrating continua are widely used to simulate gas-solid flow [14,15]. For the simulation of large-scale gas-solid fluidized system, a two-fluid model with the kinetic theory of granular flow(KTGF) initiated by Jenkins and Savage [16], Lun et al. [17]and Ding and Gidaspow [18], has become a preferred method due to its high computational efficiency [19] and successfully simulated dilute gas-solid flows [20,21]. At present, multi-size solid particles in real system has been taken into account with kinetic theory for multi-particulate flow [22]. With the coupling of population balance equation, the sub-processes such as collision, attachment and detachment in gas-solid flow have been investigated and achieved reasonable agreement with experiments [23,24]. However, the assumption that the particle-particle interactions are only binary collisions in KTGF model limits its application in dense flows [[25], [26], [27], [28]]. Therefore, the frictional stress is proposed in the solid shear viscosity in dense regions and several frictional models have been successfully used into the dense flow [[29], [30], [31], [32], [33], [34], [35], [36]].

The interphase momentum model especially the drag model has an important influence on the gas-solid flow simulation [[37], [38], [39], [40], [41], [42]]. Several average-based drag models such as Wen and Yu [43], Syamlal and O'brien [44] and Gidaspow [15], have been proposed in TFM which based on the assumption of particles distributed homogeneously in computational grids. However, some reports found that the average-based drag models overestimated the drag force [[45], [46], [47], [48]]. Li and Kwauk [49,50]found the strong dependence of the drag coefficient with mesoscale structure. Some researchers suggested that the heterogeneous flow sub-grid microstructure had significant effects on drag force [[51], [52], [53], [54], [55], [56]]. However, TFM has issues that it requires high grid resolution and small steps to capture the dynamic evolution of particle-fluid mesoscale structures [[57], [58], [59], [60]].

In order to account for the effects of mesoscale structures, the drag model based on the energy minimization multiscale model(EMMS) is incorporated into the two-fluid model to simulate gas-solid flow [61,62]. Lu et al. [[63], [64], [65]] has proven that the EMMS-based approach reaches a mesh-independent solution of the sub-grid structure and has an accurate result in gas-solid dynamic simulation for both Geldart group A, B particles. The EMMS-based drag model has been widely reported to predict gas-solid flows and the results showed a higher accuracy prediction than homogeneous drag model [[66], [67], [68], [69]].

The two-fluid model has been applied for sand shooting process and there are some publications on it. Winartomo et al. [70]proposed the multiphase model for simulation of core shooting process and a qualitatively validation was made by experiments. Wu et al. [[71], [72], [73]] applied the TFM with KTGF for the core shooting simulation. Their simulation results were coincident with the testing results. Bakhtiyarov and Overfelt [1] investigated the rheological properties of resin-bonded sand/air mixture with TFM and Ostwald model. With the numerical predictions and experimental measurements, pressure variation and flow rate were presented in the cold box core molding process. Pelzer et al. [74] took viscosity, momentum exchange and solid pressure into consideration. With special emphasis of these items on the common two-fluid model, a sufficient agreement was achieved between numerical results and experimental observations. Ni et al. [[75], [76], [77]] applied TFM with a kinetic-frictional constitutive correlation to simulate flow dynamics of core shooting process and found that the vents and sand properties have a significant influence on flow behavior of particles. Tong et al. [78] investigated the influence of restitution on the sand flow behavior by TFM and found that restitution leads to a difference of sand particles distribution and velocity. However, the sand shooting process is a dynamic transformation process with dilute, dense and gas-particle interaction phase in sand box. The heterogeneous flow sub-grid microstructure is needed to investigate in sand shooting process.

In this paper, the EMMS-based drag model which has a widely used in fluidized bed is incorporated into the TFM to investigate the sand shooting process. A grid dependency study is made to validate the grid independence of the model. Several wall boundary conditions (free-slip, partial-slip and no-slip) were introduced to investigate the effect of specularity coefficient on sand flow behavior. Experiments with transparent core box and computer controlled high-speed camera were used to study flow dynamic behavior in the sand shooting process with simulation results.

In order to study the influence of heterogeneous microstructure on flow behaviors, a comparison between EMMS-based and Gidaspow drag model was made.