MULTI-FIDELITY SURROGATE MODELS FOR WELDING APPLICATIONS Pablo Pereira Alvarez, Julien Pelamatti and Sami Hilal Keywords: Multi-fidelity, Surrogate models, Welding, Finite Element Method Welding is a widely spread manufacturing and repair technique in most industrial ac- tivities. Mastering the process and its consequences is of key importance in the strict safety environment of the nuclear industry. Numerical simulation of welding has shown its usefulness for residual stresses estimation [3]. However, it is well-known for being computationally expensive, as it deals with large multi-physics finite element models. In this context, surrogate models are a practical solution that allows for quick and accu- rate estimations of quantities of interest. The main disadvantage of surrogate models is that they need to be trained with rich parametric studies that might not be feasible in reasonable time. Multi-fidelity approaches [1] are a suitable alternative to classic methods for restrained computation budgets. The models are constructed from a mix of high fi- delity expensive simulations and low fidelity faster simulations, which allows to reduce the number of high fidelity simulations. Recent advances in multi-fidelity surrogate models [2] show promising results for non-linear relationship between the fidelities. Several multi-fidelity methods are tested on a simplified groove welding application with different definitions of fidelity: from mesh refinement to constitutive law modelling. The first results on hardness and residual stress prediction are encouraging and showcase the potential of multi-fidelity surrogate models for welding applications. REFERENCES [1] Loı̈c Brevault, Mathieu Balesdent, and Ali Hebbal. Overview of gaussian process based multi-fidelity techniques with variable relationship between fidelities, application to aerospace systems. Aerospace Science and Technology, 107:106339, 2020. [2] Kurt Cutajar, Mark Pullin, Andreas Damianou, Neil Lawrence, and Javier González. Deep gaussian processes for multi-fidelity modeling. arXiv preprint arXiv:1903.07320, 2019. [3] V Robin, S Hendili, J Delmas, S Hilal, D Iampietro, M Abbas, and S Jutteau. Modelling of residual stresses in multi-pass pipe circumferential butt welds made of austenitic stainless steel to provide indicators for scc risk classification. In Pressure Vessels and Piping Conference, volume 87486, page V005T06A098. American Society of Mechanical Engineers, 2023.