Estimation of Casting Mold Interfacial Heat Transfer Coefficient in Pressure Die Casting Process by Artificial Intelligence Methods

Abstract

Pressure casting process, which is based on the principle of filling and solidifying the liquid metal into the mold cavity with the effect of speed and pressure, enables to obtain a serial product. The pressure casting process usually involves a thermal process. Starting with the casting process, the thermal resistances, especially formed at the casting mold interface, and the resultant interfacial heat transfer coefficient (IHTC) are among the most important factors determining the mechanical and physical properties of the produced part. The IHTC depends on the mold temperature, casting temperature, injection pressure, injection rate, vacuum application and many other incalculable parameters. In this study, it was aimed to determine the heat transfer coefficient and heat flux of the casting mold interface which has a significant effect on the quality of parts in the pressure casting of cylindrical mold geometry of AlSi8Cu3Fe aluminum alloy. The study was carried out depending on different casting temperatures, injection pressure, injection speed and vacuum application to the mold cavity. Temperatures were measured with thermocouples placed in the mold and casting material, IHTC and heat flux were calculated with finite difference method by using experimentally measured temperatures. In the application of artificial intelligence methods, casting temperature, injection speed, injection pressure and vacuum conditions are given as input parameters and interfacial flow coefficient and heat flux are accepted as output parameters. With the help of these parameters, DTR, MLR and ANNR deep learning algorithms were used to estimate the interfacial heat transfer coefficient. Among these algorithms, ANNR algorithm was found to be the most accurate estimating model at the rate of 99.9%. For the obtained model, a computer program was prepared for the users to be able to see and follow the experimental results and the results obtained from the model at the same time.