Research On Temperature Field Model Of Weld Arc Additive Manufacturing

In the field of aerospace,the technology of Weld Arc Additive Manufacturing(WAAM)is increasingly being chosen for the preparation of large metal components,gradually replacing traditional manufacturing and becoming the mainstream.The high deposition rate,wide range of applicable materials,and high quality forming in both microscopic and macroscopic dimensions of WAAM are unparalleled advantages of other manufacturing processes,which solve the problems of difficult and slow processing in traditional processes.However,the drastic and rapid temperature history in WAAM process also pose problems.This kind of additive manufacturing technology uses wire arc as the heat source,and its high heat input and thermal cycle history will lead to the residual thermal stress in formed parts.Therefore,it is crucial to study the temperature evolution process of WAAM.In order to suppress defects from the root cause and achieve high-quality forming by predicting the additive temperature,this paper will focus on studying the temperature field model of the wire arc additive manufacturing process.Firstly,based on the temperature field distribution characteristics of WAAM and the numerical simulation theory of temperature field,the double ellipsoidal heat source model is selected to simulate the temperature field.With the help of Ansys Workbench,a finite element model of the temperature field is established based on the temperature control equation.Three types of boundary conditions,including temperature,heat flux density and surrounding medium temperature and heat transfer coefficient,are set up.And temperature dependent materials’ thermophysical parameters are introduced.Use command and the method of birth-and-death technology to achieve the load of moving heat source.Use infrared thermometer to monitor the whole process to verify the model,and obtain reasonable and accurate temperature field model of WAAM process.Then,the double ellipsoidal heat source model in the temperature field model is studied.And the correlation mechanism between its parameters and the temperature field distribution is specifically analyzed.An automatic extraction technology of molten pool shape is proposed.The calculated results of the temperature field model are post-processed,and the obtained molten pool boundary and sizes are used to analyze the distribution characteristics of the molten pool temperature field.By comparing the thermal cycle curves of temperature field under different thermal efficiency values,a linear regression model of peak temperature and thermal efficiency is obtained by cftool fitting.For the shape parameters of double ellipsoidal heat source,multiple linear regression analysis is used to establish regression models for the depth,width,and peak temperature of the molten pool.Provide basis for the selection of subsequent heat input and heat source model parameters.Further,experiments are carried out to investigate the distribution characteristics of the single layer and multi-layer temperature field respectively.A three factor and four level orthogonal experiment is designed to conduct a single layer experiment to obtain a stable forming process parameter range.Carry out 20-layer experiments,cut and observe the cross-section of the formed parts,measure the layer width of each layer,analyze the continuous layer width change curve and continuous eccentricity curve.Summarize the relationship between heat input and forming size of multiple layers’ additive manufacturing.The distribution of temperature field in the molten pool during the simulation of multi-layer additive process shows that as the number of additive layers increases,the average temperature gradient in the molten pool shows a decreasing trend,and the heat in the molten pool mainly transfers along the direction of additive height which lay a foundation for subsequent temperature prediction of WAAM.Finally,a neural network temperature prediction model is established and optimized in Matlab.Using the current,weld speed,feed speed,measurement points’ location information,and molded parts’ size as inputs to the neural network,and temperature as output,the neural network is reasonably designed and trained to obtain a single layer temperature prediction network with the structure of 6-9-1,and a multi-layer temperature prediction network with the structure of 7-4-1.Using genetic algorithm to optimize the neural network and improves the accuracy of the prediction model.Based on this prediction model,design a multi-layer additive process temperature prediction software,and ultimately achieved the prediction of additive temperature by inputting process parameters,model size,and location.