Improving the melting efficiency in aluminum melting furnaces has significant cost and environmental benefits. A large amount of data, including time-series data, have been accumulated from the use of aluminum melting furnaces, but no effective method has been established to utilize these data. In this study, a data-driven model was constructed by combining two machine learning methods : variational autoencoder (VAE) and artificial neural network (ANN). VAE was applied as a model to quantify time series data into 18 latent variables, while ANN was constructed as a model to predict fuel gas consumption from latent variables and other characteristics. In addition, we attempted to optimize aluminum melting process by simulation using the data-driven model.

  Although the aluminum melting process was complicated, we were able to construct a highly accurate prediction model (R² ＝ 0.69). Furthermore, the characteristics of the fuel gas flow rate in the case of high melting efficiency were determined by simulation. In fact, the results of modifying the operating conditions of melting furnace based on the knowledge obtained confirmed a significant improvement in melting efficiency. These results indicate that the data analysis method used in this study is effective for process optimization.

  In the full mold process, sudden molten metal spouting may occur during the casting process. Few research reports have been found on this dynamic phenomenon, and it is difficult to explain the phenomenon using a hot-metal flow model based on gas pressure in the pyrolysis layer. This paper reports the results of the observation of the filling behavior in molds by X-ray, which was conducted to understand this dynamic phenomenon. The cast iron used was FC250, and the model was made of PMMA material with a 65-fold foaming magnification ratio. A commercial paint mold (air permeability 1) was flow-coated over the model, and dried twice. A straight ceramic tube with an inner diameter of 25 mm was used for the mouth system, and casting was performed with a side bottom gating. Once the molten metal flows into the mold, its ball shape increases in size, and soon it begins to be violently blown from the wall opposite the gate towards the gate. Then, as the water level rises, the molten metal changes its behavior such that it protrudes out of the mold corner on the gate side. The high-speed video images show that both the blowing away of the metal surface and protrusion of molten metal are caused by the spouting of molten metal from the bottom of the mold in a 0.1-second cycle. This suggests that gas explosions occur at the bottom of the mold with a 0.1-second cycle, and that these phenomena are caused by the driving force of the swirling gas flow generated by the air blast running through the gap between the molten metal and coating mold. The gas explosion, swirling gas flow, and even the behavior of the molten metal are all phenomena occurring in the molten-mold gap, suggesting that the negative pressure generate in the explosion center immediately after the gas explosion caused air to be sucked in, creating a cycle of the formation of a coating / molten metal interface, which resulted in a series of explosions.

  Heat transfer coefficient (HTC) between the casting and the die is necessary to simulate flow and solidification behavior. It is known that HTC depends on the contact condition between the molten metal and die. In this study, by using FDM, the HTC was estimated from the temperature history of a cylindrical die and an aluminum alloy JIS AC4CH (A356). Three experimental variables were set : effect of hydrostatic pressure, presence or absence of die release agent, and temperature gradient in the die. Difference in HTC among the three experimental variables was found. It was also confirmed that HTC changed during the solidification process from the molten state to the completion of eutectic solidification. Difference in the contact state between solidifying alloy and die affects the HTC as previously reported.