Infrastructure for Multiphysics Software Integration in High Performance Computing-Aided Science and Engineering

Many of todays most challenging problems in science and engineering involve multiple, complex, coupled physical systems requiring advanced modeling techniques. These often involve combustion or other sources of energy release, fluid-structure interaction (FSI), computational fluid dynamics (CFD) or computational structural mechanics (CSM). Large-scale modeling and simulation of multiphysics problems using high performance computing (HPC) has become a crucial component of research and development in industry, academia, and the national laboratories. Compounding these technical challenges are prohibitively expensive commercial multiprocessor software licensing models. We propose to develop a multiphysics software integration infrastructure for use by the broader community interested in developing robust multiphysics applications, leveraging and generalizing our existing multiphysics infrastructure in Rocstar. We envision a resulting general open multiphysics infrastructure that will serve as a reference implementation in guiding the development of a new open standard for highly interoperable multiphysics software design and execution. Standardization and interface specifications for software integration interfaces across applications, languages, and HPC platform boundaries in the environment are necessary as the underlying algorithms and hardware on which they run evolves. This infrastructure will also significantly reduce the barrier to entry for private sector entities endeavoring to develop multiphysics capabilities that leverage the nations HPC resources. Commercial Applications and Other Benefits: Successful construction and implementation of the infrastructure in commercial settings will provide direct benefit to the public through the efficiency increases in modeling and simulation derived from the reference implementation. In addition to strictly technical benefits (i.e., streamlining of multiphysics capability development; significantly reducing the cost of development, ownership, and use of multiphysics simulation capabilities; easing use of modern massive computational resources), the national computational infrastructure (including installations in private industrial settings) is well-served by conforming to a standardized software development interface; the groundwork for the development of a huge number of marketable products that target the multiphysics simulation area; enabling a development substrate for advanced V & amp;V and UQ capabilities and accelerating multiscale studies (both temporal and spatial) by leveraging advanced computational tools under a common infrastructure.