Efficient Design Of Al-Si-Mg-Sc Alloys Driven By High-Throughput CALPHAD And Machine Learning

Due to their low density and high specific strength,Al-Si-Mg series alloys are widely used in different industries,including aviation,aerospace,automobile,and so on.Recent studies have shown that the addition of rare earth element Sc will greatly modify microstructure of Al-Si-Mg alloys,and thus improve the mechanical properties of alloys.However,no attempt to optimize the contents of Si,Mg,and Sc has been made due to the challenge in simultaneous screening over the high-dimensional composition space with limited experimental data.Therefore,Al-Si-Mg-Sc system was selected as the target of the present thesis.A novel alloy design strategy integrating high-throughput CALPHAD calculations,machine learning and key experiments was proposed to establish the quantitative relation“composition-process-microstructure-properties”of hypoeutectic Al-Si-Mg-Sc alloys,which was further employed to efficiently design the high-performance hypoeutectic Al-Si-Mg-Sc alloys for both gravity casting and additive manufacturing technologies.The major research achievements of the present thesis can be summarized as follows:（1）A universal high-throughput calculation platform with a novel machine learning accelerated distributed task management system（Malac-Distmas）was developed,and coupled with CALPHAD computing software and database management software to perform high-throughput calculation for various CALPHAD-type calculations（including Gibbs free energy,phase diagram,Scheil-Gulliver solidification simulation,thermodynamic properties,thermophysical properties,diffusion simulation,and precipitation simulation）and efficient data management.Moreover,the high-throughput calculation platform was used to design the optimal Sc content for commercial hypoeutectic Al-Si-Mg alloys and to optimize the heat treatment process.The optimized Sc content for commercial hypoeutectic Al-Si-Mg alloys is expressed:w（Sc）_optimal=0.09989×w（Si）–0.16948×w（Mg）–0.07436,the optimized solution temperature is in the range of 500 ^oC to 550 ^oC,and the optimized aging temperature is in the range of 149 ^oC to 188 ^oC.（2）A novel alloy design strategy integrating high-throughput CALPHAD calculations,machine learning and key experiments was proposed and successfully applied to efficiently design the high-performance hypoeutectic Al-Si-Mg-Sc casting alloys over high-dimensional composition space.Firstly,based on the established quantitative relation“composition-process-microstructure”,the relation“microstructure-mechanical properties”of hypoeutectic Al-Si-Mg-Sc casting alloys was acquired by the active learning technique supported by key experiments designed by CALPHAD-assisted and Bayesian optimization samplings.Then,a multi-objective optimization strategy was used to design high-performance alloys in high dimensional composition space.After a benchmark in A356-x Sc alloys,such a strategy was utilized to design the high-performance hypoeutectic Al-x Si-y Mg alloys with optimal Sc additions that were later experimentally validated.Finally,the present strategy was successfully extended to screen the optimal contents of Si,Mg,and Sc over high-dimensional hypoeutectic Al-x Si-y Mg-z Sc composition space,the composition space with maximum comprehensive mechanical property Q mainly locates in the range of 6～10 wt.%Si,0.5～0.6 wt.%Mg and 0.3～1.0 wt.%Sc.What’s more,an analysis strategy for the strengthening and toughening mechanisms was proposed by combining machine learning and experimental characterization,and used to reveal the strengthening and toughening mechanisms of the Sc-modified hypoeutectic Al-Si-Mg casting alloy.（3）The proposed alloy design strategy was further successfully extended to the efficient design of high-performance Sc-modified hypoeutectic Al-Si-（Mg）alloys under complex additive manufacturing processes.First,based on the exhaustive collection of data reported in the literature,the effect of composition and process parameters on the mechanical properties was systematically analyzed,the corresponding data cleaning criterion was proposed,and a high-quality dataset of SLMed hypoeutectic Al-Si-（Mg）alloys was established.Then,the machine learning model was used to establish the quantitative relationship of alloy“composition-process-property”,which was employed to design the composition and process parameters of high-performance SLMed hypoeutectic Al-Si-（Mg）alloy.Moreover,the designed alloy with a maximum Q of 966 MPa,which is 18.9%higher than the maximum Q（813MPa）of all the available experimental data.Finally,the effect of Sc addition on the crack sensitivity and microstructure of the hypoeutectic Al-Si-（Mg）alloys was studied.The calculated results show that Si and Sc have a noticeable influence on the crack sensitivity of the alloy:The larger the Si content,the smaller the cracking tendency.Furthermore,when the addition of Sc exceeds 0.04 wt.%,the cracking tendency of the alloy can be effectively improved.