This resource requires you to log in before you can proceed with the download.
Validated models (be they data driven or physics based) can accelerate the design, certification, and ultimate deployment of novel materials and structures by allowing rapid iteration and virtual experimentation while at the same time minimizing the need for expensive and time consuming physical experiments. Multiscale modeling is often necessary to fully capture the effects of relevant physical mechanisms spanning broad time and spatial scales that will affect performance. However, multiscale models must often balance fidelity (accurately accounting for different length and time scales) with tractability. Hierarchical multiscale models rely on precomputed homogenized properties thereby often sacrificing fidelity for computational speed while concurrent multiscale models, that explicitly incorporate microscale features at every integration point, enhance fidelity but at great computational expense. Recently, machine learning surrogate models have shown great success in accurately mimicking physics-based models with orders of magnitude reduction in relative computational speed. This lecture is intended to provide an overview of NASA Glenn’s multiscale modeling toolset (based primarily on the family of method of cells micromechanics theories) and discuss practical developments toward accurately predicting the thermomechanical behavior of composite materials and structures.
Cite this work
Researchers should cite this work as follows: