Abstract: The artificial neural network technique is introduced to calculate the locations of metal-pattern interface according to the mechanism of the molten metal filling and metal front moving during the filling process of lost foam casting (LFC). Based on the calculated interface location, the fluid flow of the molten metal during mold filling process is calculated by use of SOLA method and the temperature distribution is coupled with. The temperature distribution during the solidification process is calculated by the numerical solution of the three-dimensional heat conduction partial differential equation according to the casting and mold temperature distribution at the moment of filling process ending. The shrinkage defect of ductile iron lost foam casting is predicted according to the characteristics of mold wall movement in LFC. The results of simulation are consistent both with the results of mold filling test and with the results of casting intersecting.

Key words: Artificial neural network, Computer simulation, Lost foam casting