O.H. Griffin
Distributed computing resources for simulation and visualization are both required for development, solution, and post processing of the detailed micromechanics models used for analysis of textile composite materials. Most of the pre- and-post processing of these models can be accomplished on an Indigo II machines. The solution phase of small models that are less than 50 degrees of freedom (d.o.f.) may also he performed on a Indigo II. For models larger than 200K d.o.f., scalable midsize computer.
Scalable computing resources and visualization of results in a CAVE environment will enable the analysis of detailed micro mechanical models of textile based composites. While traditional aerospace industry over the past 20 years, textile based composites are a recent development in the area of lightweight structural materials. Textile composites (TC) have the advantage over laminated composites of significantly greater damage tolerance and resistance to delamination. Currently, the major disadvantage of TC is the inability to examine the details of the internal response of these materials under load. Although more costly than simple strength of material models, the present analysis, based on detailed finite element models of the representative volume element (RVE) of a textile, allows prediction of the load, mode, and location of failure within the RVE. Through these models, not only is gross characterization possible, but internal details of displacement, strain, stress, and failure parameters can be studied.
Preliminary analyses on an IBM RS6000 have proven useful in the study of simple textiles such as plain weaves. Since the RVE of a plain weave contains only 2 yarns, it could be modeled with 45K d.o.f. Solution of a linear elastic analysis required approximately 600Mb of disk and 100 minutes of c.p.u. time. The largest proposed models are two-dimensionally braided textiles, having ten to fifteen yams in the RVE and requiring in excess of 200K d.o.f. for fully converged solutions. These models will incorporate an anisotropic progressive failure methodology requiring multiple load steps and multiple iterations within each load step. This combination of the finite element method and progressive failure methodology should allow the numerical simulation of the response of a textile composite through final failure. Computer requirements for such a procedure should be between one and two orders of magnitude greater than that which was required for the simple plain weave model. Software requirements: IDEAS, ABACUS , AVS. This research would be significantly enhanced if these existing desktop workstation tools could also be used in a CAVE environment, particularly for complex RVE geometries and corresponding strain energy density distributions.