Finite difference methods in the regular Cartesian grid are often used for mold filling simulations. The main advantages of these methods are memory and CPU saving and ease of grid generation compared to other unstructured methods. However, representation accuracy of casting shape is very poor; for example, slopes or curves are represented as stair-steps. Therefore, in such cases, calculated results are rarely consistent with the actual phenomena. However, these disagreements are not only caused by poor shape representation but also by the numerical error of the upwind scheme. In other words, high order schemes are expected to provide more accurate solutions without improving stair-step representation of the casting shape. The aim of this work was to investigate the stair-step representation, and to improve the accuracy of numerical analysis by the CIP (Constrained Interpolation Profile) method as high order scheme. By speculating the problems of stair-step approximation, it was found that inappropriate pressures are caused by numerical deterioration of flow velocity. The decay is proportional to the n-th power of the Courant number, if the accuracy of the scheme is in the time-space n-th order. In general, the Courant number is less than 1. So the error of stair-step representation can be reduced by using the high order scheme. Some problems were solved by the upwind scheme and CIP method. In the simulated results by the upwind scheme, the error of stair-step generated strongly. On the other hand, it was found that CIP method reduces the error.

View full abstract

  Recently, many products are becoming more and more miniatured in shape because of growing demands for light-weight or cost reduction. As product size decreases, compared with gravity and inertia force, influence of surface tension increases relatively. Therefore, it is required for fluid flow simulation to take into account the effects of capillary and wetting. In this study, fluid flow simulation program based on SOLA-VOF method for surface tension dominant phenomenon was developed. The resulfs showed significant difference between the results calculated with the effect of surface tension and without it. The effect of surface tension increases with decreasing Weber number.

View full abstract

  It is important to predict shrinkage porosity quantitatively in squeeze casting process by CAE. A new calculation method has been developed considering molten metal supply by pressure in squeeze casting. Using this method, the size of shrinkage-porosity in an aluminum suspension part can be simulated quantitatively. The predicted results were compared with the X-ray inspected results of actual castings, and the size of shrinkage porosity was investigated by changing squeeze pressure. As a result, the predicted size of shrinkage porosities was found to agree well with the sizes of shrinkage porosities of the actual castings.

View full abstract

  Calculation time reduction of finite difference solidification analysis by implicit-explicit FDM combines mingles implicit and explicit methods (hereafter called the hybrid method) was investigated. The hybrid method was proposed in 1991 to reduce the calculation time of finite difference heat analysis for transient heat transfer. This paper describes the application of the hybrid method to finite difference solidification analysis. Also, the relationship between calculation time reduction and calculation accuracy was investigated. In hybrid method applied FDM solidification analysis, estimated solidification time was found to always deviate from the full-explicit results while the estimated macro shrinkage porosity did not deviate that much. The preferned results of below 1.1% solidification time deviation from full explicit calculation were obtained in this case when time step magnification was between 2 and 4.

View full abstract

  With growing needs for thin wall and weight-saving products, production quantity of aluminum alloy is increasing more and more. It is important to control casting conditions in order to obtain sound and highly functional products. The present study focused on the cold flakes in the shot sleeve of aluminum alloy die casting. The solidified layer generated in the early stages of die casting when the molten metal is charged in contact with the bottom of the sleeve, and exfoliated with the movement of the plunger, yielding the cold flakes. If the cold flakes flow into the cavity region, the mechanical properties of product will deteriorate. To predict casting defects caused by cold flakes, numerical simulation was performed to analyze the dynamic behaviors of the cold flakes. The conditions for the generation of the cold flakes and the equations of motion including gravity, buoyancy, fluid drag were established in the present study. The movement of cold flakes was traced by solving momentum equations by the Runge-Kutta-Gill method. The calculated results showed that the behavior of cold flakes vary according to conditions such as initial temperature of molten metal, sleeve temperature, etc.

View full abstract

  An automatic riser optimization system was developed to find casting designs that would provide the highest product quality at the lowest costs. Accurate and quantitative prediction of the size and shape of casting defects is indispensable for the evaluation of casting quality. A new numerical method was proposed based on simplified molten metal flow due to volume change during solidification. The calculated macroshrinkage volumes and distribution of microshrinkage agreed well with the experimental observations and X-Ray CT inspections in Al alloys castings.
  By using the proposed prediction method, a CAO framework was developed for automatic prediction of optimum casting design without shrinkage cavity in products. T-shaped steel casting was submitted to the CAO framework to find the diameter and height of the riser that world minimize the riser volume while producing defects-free products. The obtained parameters by the CAO were consistent with the results of past experiments. This comparison shows that the developed optimization techniques are applicable for optimizing automatic casting design and thus contribute greatly to the improvement of yield and quality.

View full abstract

  Casting design to achieve good metallurgical bonding in the cast-in insertion process was analyzed using flow and solidification simulation, and was verified experimentally. For steel pipe insertion in a cylindrical cast iron, and for bonding a steel ring to an end face of a hollow cylinder, eight specimens were cast by various systems with different gates, risers, run-offs, and additional thicknesses. The enhancing effect of Ni alloy plating on the inserts was also examined. Cast specimens were cut up and the bonding state was evaluated. The experiments showed that every modified casting system was effective for improving bonding even in un-plated specimens. In Ni plated specimens, all of the tested systems achieved good bonding although some small defects were observed on or near the interface depending on the position in the castings. Comparing the results of experiments and the flow and solidification simulation, it was concluded that smooth melt flow without turbulence or stagnation, and immediate contact of the interface with the melt in pouring are required for good bonding by minimizing the interface oxidization. The Ni plating helps prevent oxidation and shorten the time required for liquid melt contact with the interface at temperature equivalence, resulting in reduced sensitivity to thermal conditions at the interface.

View full abstract

  Fluidity of flake graphite cast iron was experimentally determined using silica tube as the fluidity channel and suction by a regulated vacuum system. A simulation program was developed to predict fluidity length and solidification structure of fluidity specimens. Flow velocity was calculated by the modified Bernoulli's equation taking into consideration viscosity loss and surface tension loss. From both the experiments and simulation, a linear relation between fluidity length and tube diameter was obtained. Fluidity length increased with increasing suction pressure and melt temperature. Fluidity length was linearly related to the square root of suction pressure. The microstructure of a fluidity specimen consists of three regions from the tip backward: ledeburite, motile and flake graphite structure. The cooling rate near the end of solidification at different positions along the fluidity specimen length was calculated and successfully correlated with different structures, assuming the critical cooling rates for each structure.

View full abstract

  The fluidity of AC4CH aluminum alloy was experimentally determined using silica tube, copper tube and steel tube as the fluidity channel and suction by vacuum fluidity equipment. A simulation program was developed to predict the length of fluidity specimens. Flow velocity was calculated by the modified Bernoulli's equation taking into consideration viscosity loss and surface tension loss. From both the experiments and simulation, the calculation results corresponded with the experiment value assuming the critical fraction of solid to be 0.35. The flow length decreased in order of quartz tube, carbon steel tube, and copper tube. This order agreed with the results of comparing the calculated total thermal resistance. The fluidity improved by the combined use of coating and argon atmosphere. By comparing the simulated results with the experimental results, the flow length in the test tube with the coating was found to increase because the coefficient of heat transfer decreases due to insulating effects.

View full abstract

  Although numerous works on numerical simulation of mold filling have been reported, their accuracy is not clear because of the difficulty of direct observation of mold filling. This paper presents the direct observation results of mold filling and comparison with simulation. In direct observation, X-ray absorption images of cast iron melt flow in sand molds were recorded with a high-speed digital video camera. In particular, the effects of gating systems on the filling behavior into cavities were investigated. Numerical simulation was carried out with a code where the governing discrete equations for the momentum, mass and energy conservation laws have been derived by the DFDM (Direct-Finite-Difference-Method). The observation results revealed the following; although numerical simulation under same conditions results in the same mold filling, real casting does not, in particular on the entrapment of gas bubbles; considerable gas was entrapped during mold filling and most of the gas disappeared with the melt flow; considerably smaller sprue cross-section and higher back pressure can realize much smoother mold filling without gas trapping; fins are formed when the mold was filled completely, etc. Although computer simulation results roughly agreed with observed ones, it was difficult to well simulate flow separation on curved surfaces, detailed wavy free surface, and flow change in similar pouring.

View full abstract