In this paper we address the particular need for high-speed or “real-time” characterization of realistic anisotropic material systems such as laminated composites. This is driven by the desire to dynamically alter the loading paths applied by a multiaxial robotic test frame during the testing of a specimen, so that strain states are developed in the specimen in a manner that activates the maximum excitation of the specimen’s constitutive properties. In order to achieve this goal, we present an evolutionary adaptation of earlier work into computationally efficient material characterization using response-surface surrogate models. This approach is enhanced by the adoption of highly-parallel General Purpose Graphics Processing (GPGPU) computing. We discuss the challenges of adapting the characterization problem for GPGPU computing, particularly in terms of parallelization, synchronization, and approximation. Two parallelized algorithms for characterization are developed, and the merits of each are discussed. We then demonstrate validation results on a simple linear-elastic material system, and present statistical data which demonstrate the robustness of the approach in the presence of experimental noise. We conclude with remarks regarding the performance of the GPGPU-enabled characterization algorithm, and its applicability to more complex material systems.

This content is only available via PDF.
You do not currently have access to this content.