Abstract:
In order to study the formation of the solidification structure including the columnar-toequiaxedtransition (e.g., CET) under the influence of ultrasound, a 2-zone furnace system and an ultrasonic equipment were built and utilized. The 2-zone furnace system consists of an induction furnace with a top coil and a bottom coil, a graphite crucible, and. a water-cooled chill block located at the bottom of the crucible. By controlling both the top and bottom coil output power independently, the furnace can create various temperature gradients and cooling rates in different regions of the graphite crucible. The ultrasound Nb probe was inserted at the top of the crucible. The top of the crucible was also thermally insulated. Temperature measurements were performed at different locations in the crucible. The effects of ultrasound on the microstructure formation during solidification of A356 alloy was studied. A numerical model was developed to simulate the solidification process in the crucible and to assist in developing of solidification maps. In addition, with the help of machine learning techniques, the mutual interaction between Al-Si based alloys content, physical properties and processing parameters have been studied. Quantitative relationship equations and the significance of each key factors were developed and analyzed via machine learning, which helps to better understand the complex nonlinear relationship of “alloy composition-physical properties-solidification processing parameters-mechanical performance” in cast Al-Si based alloys. In the first chapter, the background of A356 alloy, ultrasonic stirring technology and the application of machine learning technique on metallurgical and materials engineering are introduced. The second chapter describes the methodology of the ultrasonic refining and modification mechanisms as well as several relevant machine learning algorithms that can be applied to study metallic materials. In the third chapter, the ultrasound effects on the formation of the solidification structure of A356 ingots processed via a 2-Zone induction melting furnace are studied. The fourth chapter focuses on modeling of segregation and microstructure evolution during the solidification of A356. Finally, in the fifth chapter a prediction of a quantitative relationship of “alloys content-physical properties-processing parameters-mechanical properties” using machine learning is performed.
