17th AIAA Computational Fluid Dynamics Conference, June 43rd AIAA Aerospace Sciences Meeting and Exhibit, June 6–9, 2005/Toronto, ON Multi-Code Simulations: A Generalized Coupling Approach J. U. Schl¨ uter ∗ , X. Wu † , E. v. d. Weide ‡ , S. Hahn § , J. J. Alonso ¶ , and H. Pitsch ‖ Center for Turbulence Research & Aerospace Computing Lab Stanford University, Stanford, CA We present an approach to perform multi-physics simulations by coupling multiple solvers. In order to simplify the coupling process, we have developed the software module CHIMPS (Coupler for High-performance Integrated Multi-Physics Simulations) that han- dles the coupling of two mesh-based solvers. The module is based on the script language python and has been developed to accommodate a wide variety of mesh based solvers. In this paper we will present previous multi-code approaches, the advantages of the new approach using the software module CHIMPS, and present some verification test-cases. I. Introduction S imulations of real world engineering applications require that a large variety of physical phenomena need to be modeled. The area of application for numerical simulations have been extended significantly in the recent years due to the expansion of computational resources. However, the integration of new models into existing solvers is often problematic. Let us consider the example of fluid-structure interactions. In the past, flow solvers have been developed and optimized to simulate a variety of flows. Solvers for structural analysis have been developed in parallel to simulate stresses and deformations in solid bodies. So far, in order to simulate fluid-structure interactions, both approaches have to be integrated into a single solver. In the process, usually one had to accept that one of the models was not as accurate, optimized, or flexible as in the original solver. Another example is that of LES-RANS Hybrids. Here, a variety of flow solvers have been written for either the RANS (Reynolds-Averaged Navier-Stokes) approach, or LES (Large-Eddy Simulations). While previous LES-RANS hybrid approaches, such as Detached-Eddy Simulations (DES) 1 and Limited-Numerical Scales (LNS) 2 combine LES and RANS in a single flow solver, the approach to couple two existing flow solvers has the distinct advantage to build upon the experience and validation that has been put into the individual codes during their development, and also to run simulations in different domains at different time-steps. Both approaches, LES and RANS, require a distinct set of algorithms and models to work efficiently and accurately. The integration of both approaches into a single solver is often tiresome. Here, we want to present a different approach to address this problem: We keep the solvers separate. The solvers run simultaneously and exchange only the necessary information at the interfaces. This allows that each solver is using the best and most accurate methods for the solution of its problem, while the interaction with other solvers allows to take on very complex engineering applications. In the example of fluid-structure interaction, we can execute one flow solver and one structural solver simultaneously. At the surface of the body the structural solver receives the pressure loads from the flow solver and the flow solver receives the displacement of the body from the structural solver. ∗ Research Associate † Research Associate ‡ Research Associate § Postdoctoral Fellow ¶ Associate Professor ‖ Assistant Professor Copyright c  2005 by Center for Turbulence Research, Stanford University. Published by the American Institute of Aeronautics and Astronautics, Inc. with permission. 1 of 12 American Institute of Aeronautics and Astronautics Paper 2005-4997