Survey of Sensitivity Analysis Methods During the Simulation of Residual Stresses in Simple Composite Structures

Citations Over Time

Abstract

Process-induced residual stresses occur in composite structures composed of dissimilar materials. As these residual stresses could result in fracture, their consideration when designing composite parts is necessary. However, the experimental determination of residual stresses in prototype parts can be time and cost prohibitive. Alternatively, it is possible for computational tools to predict potential residual stresses. Therefore, a process modeling methodology was developed and implemented into Sandia National Laboratories’ SIERRA/Solid Mechanics code. This method requires the specification of many model parameters to form accurate predictions. These parameters, which are related to the mechanical and thermal behaviors of the modeled composite material, can be determined experimentally, but at a potentially prohibitive cost. Furthermore, depending upon a composite part’s specific geometric and manufacturing process details, it is possible that certain model parameters may have an insignificant effect on the simulated prediction. Therefore, to streamline the material characterization process, formal parameter sensitivity studies can be applied to determine which of the required input parameters are truly relevant to the simulated prediction. Then, only those model parameters found to be critical will require rigorous experimental characterization. Numerous sensitivity analysis methods exist in the literature, each offering specific strengths and weaknesses. Therefore, the objective of this study is to compare the performance of several accepted sensitivity analysis methods during the simulation of a bi-material composite strip’s manufacturing process. The examined sensitivity analysis methods include both simple techniques, such Monte Carlo and Latin Hypercube sampling, as well as more sophisticated approaches, such as the determination of Sobol indices via a polynomial chaos expansion or a Gaussian process. The relative computational cost and critical parameter list are assessed for each of the examined methods and conclusions are drawn regarding the ideal sensitivity analysis approach for future residual stress investigations.

Related Papers

• → Efficient sampling algorithm for large-scale optimization under uncertainty problems(2018)56 cited
• → Identification of stochastic interactions in nonlinear models of structural mechanics(2017)9 cited
• → An Arbitrary Polynomial Chaos-Based Approach to Analyzing the Impacts of Design Parameters on Evacuation Time under Uncertainty(2014)7 cited
• → IPSAL: Implementation of the module to generate the Sobol sequence and indices(2023)3 cited
• Efficient Sampling Algorithm for Optimization Under Uncertainty(2016)