Python Vectorized User MATerial (PyVUMAT) Model 1.0 User’s Guide

Author(s):

Joshua C. Crone

Abstract:

The Python Vectorized User MATerial (PyVUMAT) framework enables users to develop custom material models in Python and seamlessly integrate them into finite element analysis simulations. Development in Python can reduce development time significantly compared with compiled languages (i.e., C, C++, and Fortran) by leveraging the vast collection of freely available Python packages. This is especially true for the emerging field of machine learning (ML)-based material models, where established Python packages such as PyTorch, TensorFlow, and JAX can drastically simplify ML model development. Direct use of these packages will allow ML material model development to keep pace with the rapid advancements in the ML community. PyVUMAT may also lower the barrier to creating novel material models for researchers who are unfamiliar with C, C++, or Fortran. The first half of this user’s guide provides the necessary information to start developing and running PyVUMAT models. The latter sections provide additional guidance on improving functionality and performance.