Livermore Software Technology Corporation

Willem J Roux

Tushar Goel

Improved topology optimization for engineering product design is disclosed. An engineering product including a design domain to be optimized is defined. Design domain can be a portion or the entire engineering product. Design objective and optional constraint are defined such that optimization goal is achieved. Additionally, initial configuration of the design domain is represented by a finite element analysis (FEA) mesh. Each element or element group is associated with a design variable. A set of discrete material models is created from the baseline material used for the design domain. The set of discrete material models is configured to cover entire range of the design variable and each discrete material model represents a non-overlapping portion. Each element representing the design domain is associated with an appropriate discrete material model according to the design variable. Structure response is obtained via FEA to evaluate design objective and update design variable.

Topology optimization for designing engineering product

A system, method, and software product for an accelerated design optimization is described. Engineers/designers/users define an initial design with a set of responses and constraints on the responses in a design optimization process. A plurality of approximations to the responses around the initial design is created to represent an engineering design space. A current optimal design is determined from the approximations subject to the set of constraints. Actual structural response of the current optimal design is calculated. As a result, the residual between the actual response and the approximated response can be established. When the optimization process has not reached convergence, another set of approximations is created around the current optimal design. Instead of using the original set of constraints, the new set of constraints is adjusted with the calculated residual, so that the design optimization process can achieve convergence much faster.

Accelerated design optimization

A method, a software product and a system for distinguishing effects due to bifurcation from effects due to design variable changes in finite element analysis is disclosed. According to one aspect of the invention, 1) a plurality of design experiments is analyzed with finite element analysis (FEA) software; 2) a metamodel is constructed from the FEA responses using the least squares fitting technique; 3) any FEA response that is not predicted by the metamodel is classified as outlier, which is the high likelihood candidate for bifurcation; and 4) verification of the bifurcation is then to be confirmed. The method is implemented in a design and probabilistic analysis software product.

Growing community of inventors

Livermore, CA, United States of America

Willem J Roux