Optimisation of interlayer temperature in wire-arc additive manufacturing process using NURBS-based metamodel

Abstract

For wire arc additive manufacturing (WAAM) process the interlayer temperature highly influences the quality of manufactured parts. This paper proposes an optimisation of deposition parameters for a better control of interlayer temperature while reducing the printing time employing a Finite Element (FE) model and a metamodel based on Non Uniform Rational Basis Splines (NURBS) entities for a thin-walled part in aluminium alloy. Firstly, the thermal FE model is created to extract the interlayer temperature as a function of different deposition parameters that will be optimised. These parameters are the wire feed speed and the cooling time between deposition of two consecutive layers. Then, a NURBS-based metamodel is generated to approximate the (unknown) transfer function between input variables and output responses of the problem at hand. One of the advantages of this metamodeling strategy is the possibility of obtaining the gradient of the output responses without the requirement of further computational resources, as the resulting metamodel is available in analytical form with the requisite continuity and differentiability. The NURBS-based metamodel is generated as a solution of a three-step optimisation strategy aiming at determining all the parameters defining the shape of the NURBS entity. Finally, the NURBS-based metamodel is included in the optimisation process related to the considered application. The optimisation problem is defined as a weighted sum of different criteria, i.e., total printing time and the average interlayer temperature difference for each layer. The solution obtained is subsequently validated a posteriori using the high-fidelity FE model, demonstrating an excellent agreement between the prediction of the NURBS-based metamodel and those of the FE model.