OOF: Finite Element Analysis of Microstructures
The microstructure of a material is the (usually) complex ensemble of polycrystalline grains, second phases, cracks, pores, and other features occuring on length scales large compared to an atomic size. Materials scientists are usually presented with a microstructure in the form of an image obtained from a microscope. ppm2oof is a tool to superimpose material properties onto such an image and turn it into a data representation on which oof can perform virtual experiments.
ppm2oof operates by letting the user identify regions of the microstructure with a mouse and provides various algorithms for grouping parts of an image together. ppm2oof also provides "fill-out" tables for material property specification. For instance, crystal symmetries, thermoelastic coefficients, and damage criteria can be assigned to selected regions.
ppm2oof also superimposes a computational mesh on the image and the materials data. The simplest meshing algorithm divides each image pixel into two triangular elements. This simple approach is indicative of the OOF philosophy which is to resolve an image down to a microscopic scale where material properties can be approximated as homogeneous. The degree to which a material microstructure is homogeneous depends, of course, on the magnification and resolution of the microstructural image. For example, fiberglass (a composite material) can be considered to be homogeneous for predictions of the dynamic deflection of a pole during a pole vault. On the other hand, the propagation of a microscopic crack in the pole may well depend on the fine details of the spatial correlations of the crack with the material consituents.
The simple meshing scheme of subdividing each pixel along one of its diagonals into two triangles may result in an enormous number of elements for moderately sized images and is generally memory, storage, and computationaly expensive. Adaptive meshing is available in ppm2oof and can be implemented several ways and through interaction with the user. [3] The mesh that is created represents the user's compromise between the number of elements, the homogeneiety of each element with respect to its underlying pixel properties, and the production of equilaterial triangular elements. [4]
Ultimately, the goal of ppm2oof is the creation of a file readable by oof. [5] By convention such files are called dot-goof (file.goof) files.
[3] Because ppm2oof and oof are distinct programs, no current direct capability exists that adapts the mesh to the solution. However, because one may use more than one image at a time in ppm2oof, an indirect method exists: the image-based output of an oof calculation can be used in the set of input images for ppm2oof.
[4] i.e., the dependability of the numerical interpolation provided by the elements.
[5] Or by a translator to a format for other FEM solvers.
