GENOA is an advanced, completely integrated, hierarchical software package for computationally simulating the thermal and mechanical responses of high-temperature composite materials and of structures made of those materials. The development of GENOA was guided partly by the need for a computational tool that could accelerate the design process while making it possible to avoid designing structures to be unnecessarily heavy and expensive, as they can be when one follows a deterministic approach and uses simple safety factors to account for variability among structural components in the effort to design conservatively. GENOA implements a probabilistic approach in which design criteria and objectives are based on quantified reliability targets that are consistent with the inherently stochastic nature of the properties of materials and structures.

Analytical Components Are Integrated in an iterative computational procedure with hierarchical multiple-level parallelism.

The probabilistic approach to design involves the use of the deterministic basic equations of mechanics in a more-comprehensive analysis in which stochasticity is quantified by use of probability distributions. The methods used to simulate numerically the consequences of probability distributions include the Monte Carlo method and such algebraic methods as first-order reliability, second-order reliability, the mean-value method, or the response-surface method. Probabilistic methods provide measures of the variation of risks of failure with variations in design parameters and properties of materials, thereby making it possible to determine the robustness of a design. Realistic indications of the lifetimes of structures can be obtained by taking account of such phenomena as low cycle fatigue, cracking induced by flaws, yield and ultimate strengths, creep strength, the operational environment, and damage in service. The economic benefits of using probabilistic methods to design and analyze structures include (1) reduction of weights (and thus reduction of initial costs) of structures, (2) reduction of operating and service costs, (3) reduction of failure rates, and (4) capability to generate predictable maintenance schedules.

GENOA provides computational integration and parallel processing for probabilistic mathematics and for mathematical models of composite materials and structures. Massively parallel processing enables GENOA to function in the face of the inherent complexities of high-temperature composite-material structures. Dynamic load-balancing optimization techniques are used in GENOA to minimize processing time. To perform a given analysis, GENOA takes about 1/20 of the processing time of a typical older serial-processing program developed for the same purpose.

GENOA features a highly modular architecture that makes it fast, accurate, and user-friendly. Hierarchical analytical components are implemented by software modules that contain highly specialized analysis codes (including nonlinear finite-element and micromechanical-analysis codes, for example). These components are integrated in a computational procedure that involves iteration between microscopic and macroscopic scales (see figure). The integration is effected by use of a graphical user interface (GUI) and an executive controller system (ECS). The menu-driven ECS connects the modules. The GUI provides a seamless transition from description of a problem through implementation of the solution process to post-processing graphical display of solution data. The value of integration cannot be over-emphasized: in GENOA, it is easy to import data from another structural-analysis or computer-aided-design program to describe a problem, whereas in most finite-element-analysis programs, such importation is difficult.

Analytical performance is enhanced by a capability to size adjacent problem domains dynamically to minimize processor waiting times. Central-processing-unit time is reduced and memory limitations are overcome by introduction of an effective optimized parallelization algorithm characterized by machine-independent multiple-instruction/multiple-data (MIMD), single-instruction/multiple-data (SIMD), and Open Software Foundation (OSF) types of computer architecture. Hierarchical stochastic simulation is performed to accommodate the numerous levels of uncertainty present in environmentally dependent properties of materials, enabling the user to identify quickly the most probable critical point of a design.

This work was done by Frank Abdi, Yvgeniy Mirvis, Jafar Hadian, and Kenneth J. Newell of Alpha STAR Corp. for Lewis Research Center.

Inquiries concerning rights for the commercial use of this invention should be addressed to

NASA Lewis Research Center
Commercial Technology Office
Attn: Tech Brief Patent Status
Mail Stop 7 - 3
21000 Brookpark Road
Cleveland
Ohio 44135

Refer to LEW-16543.