Computational Manufacturing: Toward Simulating Metal Casting and Welding Process

Doug Kothe, Los Alamos National Laboratory

Photo of Doug Kothe

Over the past several years, a team of researchers at the Los Alamos National Laboratory has been working on a important applications-driven project (known as “Telluride”) that endeavors to deliver a verified and validated simulation tool for particular manufacturing processes in the United States Department of Energy complex that take specialized alloys through their melting point. Gravity casting and laser/e-beam welding operations are examples of key processes that must be simulated accurately. The parallel simulation tool that represents the principal product of this project, known as “Truchas”, must execute on high-performance computational platforms and incorporate the latest physical models, linear and nonlinear solution methodology, and high-resolution PDE discretization methods. The simulation tool must accurately model the entire metal alloy casting process in one integrated simulation; hence it must embody realistic physical models for alloy solidification, micro/macro segregation phenomena, free surface incompressible flow, heat transfer, complex 3-D geometric effects, micro-structural nucleation and evolution, solid-state transition effects, and residual stress buildup and response.

I will briefly review the current capabilities and future plans for physical models and numerical algorithms embodied in the Truchas software, using high performance (parallel) simulation results where possible. Various computational and computer science issues we have faced in this project, some of which we have addressed and some of which we have not yet addressed, will be discussed.

Abstract Author(s): Douglas B. Kothe