Multifactorial Mechanism And Neural Network Prediction Of Dynamic Fracture Mechanical Properties Of Nickel-based Superalloys

As a material with excellent properties such as high temperature strength,good corrosion resistance and thermal shock resistance,nickel-based superalloys are widely used in critical components such as aerospace turbine blades and engines.However,nickel-based superalloys components are prone to damage or fracture behavior under complex and extreme service environments,but their fracture failure behavior is a complex multi-scale problem involving microscopic,mesoscopic and macroscopic,leading to a lack of in-depth and systematic studies on their damage and fracture mechanisms.Therefore,for the multi-scale fracture problem of nickel-based superalloys with different volume fractions of strengthening phases under impact loading,a multi-scale model of its dynamic fracture failure is established by molecular dynamics（MD）method and cohesive finite element method（CFEM）,and the fracture process and macro-microscopic action mechanism were systematically investigated from microscopic scale to macroscopic scale.In addition,to solve the drawbacks of traditional simulation and experimental methods with high workload and time cost,a dynamic fracture toughness prediction model for nickel-based superalloys is established based on the results of multiscale numerical simulations combined with artificial neural network methods to predict the dynamic fracture toughness efficiently and accurately.The main research work of this paper includes:（1）At the microscopic scale,the molecular dynamics method is used to reveal the influence law and mechanism of action of the reinforcing phase volume fraction and distribution,temperature,loading strain rate,on the crack extension and fracture behavior of nickel-based high-temperature alloy.When the volume fraction of reinforcing phase increases from 5%to 30%,the yield strength slowly decreases from 13.8 GPa to 13 GPa,and the average flow stress remains stable after the 15%volume fraction of reinforcing phase decreases to 7.2 GPa,and the crack extension appears through the crystal at higher volume fraction of reinforcing phase.This is due to the fact that the stress concentration near the crack front decreases as the volume fraction of the reinforcing phase increases and the initial dislocations formed around the reinforcing phase inhibit dislocation nucleation and emission within the matrix.The reinforcing phase distribution has a small effect on the yield strength of nickel-based superalloys,but has a large effect on the properties of the material during the plastic deformation phase.Due to the large mismatch between the reinforced phase and the matrix phase,whose phase interface is prone to stress concentration and formation of microporosity,leading to crack expansion along the phase boundary.As the temperature increases from 300 K to 1200 K,the Young’s modulus decreases from 139.4 GPa to 113.4 GPa,and the yield strength decreases from 12.9 GPa to9.6 GPa.At 600 K,the average flow stress reaches a peak of 7.9 GPa and then slowly decreases to 7.4 GPa,and the crack extension behavior changes from brittle crack extension to ductile extension.The dislocation line density decreases with increasing temperature during the deformation process.This is caused by the increased thermal activation effect of atoms due to the increase in temperature.The Young’s modulus increases from 139.4 GPa to 160.7 GPa and the yield strength increases from 12.9 GPa to15.7 GPa as the strain rate increases from 1.4×10¹⁰s^-1to 2.8×10¹⁰s^-1.This is caused by a significant increase in the stress concentration at the crack tip and a significant increase in the rate of dislocation proliferation under the effect of high strain rate.（2）A multiscale model of the dynamic fracture behavior of nickel-based superalloys is established,and based on this model,the influence law and mechanism of action of phase content on the fracture behavior of nickel-based superalloys were investigated.The multi-scale dynamic fracture model of nickel-based superalloys containing different volume fractions of reinforcing phases is developed by combining the molecular dynamics method（MD）and the cohesive finite element method（CFEM）with the tension-separation（T-S）curves obtained at the microscopic scale.The dynamic fracture toughness increased from 97.7 MPa·m^1/2to 115.8 MPa·m^1/2when the reinforcement volume fraction increased from 5%to 25%at a loading speed of 20 m/s.This is due to the fact that the number and distribution of the reinforcing phase become more uniform when the volume fraction of the reinforcing phase is high,which can effectively prevent the crack expansion,and the crack passes through the inner part of the reinforcing phase when the volume fraction of the reinforcing phase is higher than 15%,showing the grain penetration fracture characteristics.The reliability of the multiscale model is demonstrated by conducting three-point bending experiments for nickel-based superalloys with 15%and 20%reinforced phase volume fractions to obtain their corresponding dynamic fracture toughness,with an average error of 9%when compared with the dynamic fracture toughness from finite element simulations.（3）In order to predict and optimize the key parameters such as loading speed（15～40m/s）and reinforcement phase volume fraction（5%～25%）on the fracture toughness of nickel-based superalloys,a prediction model of dynamic fracture toughness of nickel-based superalloys is established based on artificial neural networks,combined with dynamic fracture toughness data obtained from multi-scale computational models.The network structure of the artificial neural network is adaptively optimized by using exhaustive method and hand search method to determine the best network structure.The effect of the number of samples on the performance of the network is also investigated,and it is found that the prediction accuracy and computational efficiency of the model were best when the number of samples exceeded 90.The reliability of the model is further demonstrated by performing"inferential"and"extrapolative"validation calculations on the prediction model.Finally,the model predicts that the dynamic fracture toughness of nickel-based superalloys can be guaranteed up to 90 MPa·m^1/2under complex loading conditions when the volume fraction of the reinforced phase is above 30%.