This paper describes the development of a multiphysics welding simulation model based on the discontinuous Galerkin (DG) finite-element method. Our numerical model implements a classical enthalpy-porosity constitutive law accounting for hydrodynamic and thermal effects occurring during the phase transition from solid to liquid metal. The objective of the study is to present the verification of our numerical framework and explore the applicability of the DG formulation to the simulation of welding processes. Three computational examples of increasing complexity are presented.

The design of turbomachinery relies on inexpensive, steady-state Reynolds Averaged Navier-Stokes (RANS) predictions. However, turbomachinery flows are complex with various phenomena that can be difficult to predict with RANS. Large Eddy Simulations (LES), resolving the larger scales of turbulence, appear as an attractive alternative. However, correctly resolving turbulence is challenging as numerical schemes with the correct diffusion and dispersion properties are required. The main numerical strategy employed is the Finite Volume approach on structured or unstructured grids. More recently, Finite Element-like methods, intrinsically extending up to very high orders, such as the Discontinuous Galerkin, have emerged. However, few comparisons have been made between these approaches to assess their impact on the LES predictions of turbomachinery. In this work, wall-resolved LES of a realistic linear compressor cascade is performed using three solvers, each employing one of the strategies mentionned. The results highlight that while all three approaches predict similar overall aerodynamic losses, differences in the transition mechanisms and in the turbulent kinetic energy levels are revealed. NOMENCLATURE