Michael D. Shields

The generalization of Latin hypercube sampling

Latin hypercube sampling (LHS) is generalized in terms of a spectrum of stratified sampling (SS) designs referred to as partially stratified sample (PSS) designs. True SS and LHS are shown to represent the extremes of the PSS spectrum. The variance of PSS estimates is derived along with some asymptotic properties. PSS designs are shown to reduce variance associated with variable interactions, whereas LHS reduces variance associated with main effects. Challenges associated with the use of PSS designs and their limitations are discussed. To overcome these challenges, the PSS method is coupled with a new method called Latinized stratified sampling (LSS) that produces sample sets that are simultaneously SS and LHS. The LSS method is equivalent to an Orthogonal Array based LHS under certain conditions but is easier to obtain. Utilizing an LSS on the subspaces of a PSS provides a sampling strategy that reduces variance associated with both main effects and variable interactions and can be designed specially to minimize variance for a given problem. Several high-dimensional numerical examples highlight the strengths and limitations of the method. The Latinized partially stratified sampling method is then applied to identify the best sample strategy for uncertainty quantification on a plate buckling problem.

Refined Stratified Sampling for efficient Monte Carlo based uncertainty quantification

Abstract A general adaptive approach rooted in stratified sampling (SS) is proposed for sample-based uncertainty quantification (UQ). To motivate its use in this context the space-filling, orthogonality, and projective properties of SS are compared with simple random sampling and Latin hypercube sampling (LHS). SS is demonstrated to provide attractive properties for certain classes of problems. The proposed approach, Refined Stratified Sampling (RSS), capitalizes on these properties through an adaptive process that adds samples sequentially by dividing the existing subspaces of a stratified design. RSS is proven to reduce variance compared to traditional stratified sample extension methods while providing comparable or enhanced variance reduction when compared to sample size extension methods for LHS – which do not afford the same degree of flexibility to facilitate a truly adaptive UQ process. An initial investigation of optimal stratification is presented and motivates the potential for major advances in variance reduction through optimally designed RSS. Potential paths for extension of the method to high dimension are discussed. Two examples are provided. The first involves UQ for a low dimensional function where convergence is evaluated analytically. The second presents a study to asses the response variability of a floating structure to an underwater shock.

Coarse graining atomistic simulations of plastically deforming amorphous solids

The primary mode of failure in disordered solids results from the formation and persistence of highly localized regions of large plastic strains known as shear bands. Continuum-level field theories capable of predicting this mechanical response rely upon an accurate representation of the initial and evolving states of the amorphous structure. We perform molecular dynamics simulations of a metallic glass and propose a methodology for coarse graining discrete, atomistic quantities, such as the potential energies of the elemental constituents. A strain criterion is established and used to distinguish the coarse-grained degrees-of-freedom inside the emerging shear band from those of the surrounding material. A signal-to-noise ratio provides a means of evaluating the strength of the signal of the shear band as a function of the coarse graining. Finally, we investigate the effect of different coarse graining length scales by comparing a two-dimensional, numerical implementation of the effective-temperature description in the shear transformation zone (STZ) theory with direct molecular dynamics simulations. These comparisons indicate the coarse graining length scale has a lower bound, above which there is a high level of agreement between the atomistics and the STZ theory, and below which the concept of effective temperature breaks down.

Mapping model validation metrics to subject matter expert scores for model adequacy assessment

This paper develops a novel approach to incorporate the contributions of both quantitative validation metrics and qualitative subject matter expert (SME) evaluation criteria in model validation assessment. The relationship between validation metrics (input) and SME scores (output) is formulated as a classification problem, and a probabilistic neural network (PNN) is constructed to execute this mapping. Establishing PNN classifiers for a wide variety of combinations of validation metrics allows for a quantitative comparison of validation metric performance in representing SME judgment. An advantage to this approach is that it semi-automates the model validation process and subsequently is capable of incorporating the contributions of large data sets of disparate response quantities of interest in model validation assessment. The effectiveness of this approach is demonstrated on a complex real-world problem involving the shock qualification testing of a floating shock platform. A data set of experimental and simulated pairs of time history comparisons along with associated SME scores and computed validation metrics is obtained and utilized to construct the PNN classifiers through K-fold cross validation. A wide range of validation metrics for time history comparisons is considered including feature-specific metrics (phase and magnitude error), a frequency metric (shock response spectra), a time-frequency metric (wavelet decomposition), and a global metric (index of agreement). The PNN classifiers constructed using a Parzen kernel for the class conditional probability density function whose smoothing parameter is optimized using a genetic algorithm performs well in representing SME judgment.

Adaptive Monte Carlo analysis for strongly nonlinear stochastic systems

Abstract This paper compares space-filling and importance sampling (IS)-based Monte Carlo sample designs with those derived for optimality in the error of stratified statistical estimators. Space-filling designs are shown to be optimal for systems whose response depends linearly on the input random variables. They are, however, shown to be far from optimal when the system is nonlinear. To achieve optimality, it is shown that samples should be placed densely in regions of large variation (sparsely in regions of small variation). This notion is shown to be subtly, but importantly, different from other non-space-filling designs, particularly IS. To achieve near-optimal sample designs, the adaptive Gradient Enhanced Refined Stratified Sampling (GE-RSS) is proposed that sequentially refines the probability space in accordance with stratified sampling. The space is refined according to the estimated local variance of the system computed from gradients using a surrogate model. The method significantly reduces the error in stratified Monte Carlo estimators for strongly nonlinear systems, outperforms both space-filling methods and IS-based methods, and is simple to implement. Numerical examples on strongly nonlinear systems illustrate the improvement over space-filling and IS designs. The method is applied to study the probability of shear band formation in a bulk metallic glass.

Stochastic collocation approach with adaptive mesh refinement for parametric uncertainty analysis

Abstract The presence of a high-dimensional stochastic input domain with discontinuities poses major computational challenges in analyzing and quantifying the effects of the uncertainties in a physical system. In this paper, we propose a stochastic collocation method with adaptive mesh refinement (SCAMR) to deal with high dimensional stochastic systems with discontinuities. Specifically, the proposed approach uses generalized polynomial chaos (gPC) expansion with Legendre polynomial basis and solves for the gPC coefficients using the least squares method. It also implements an adaptive mesh (element) refinement strategy which checks for abrupt variations in the output based on a low-order gPC approximation error to track discontinuities or non-smoothness. In addition, the proposed method involves a criterion for checking possible dimensionality reduction and consequently, the decomposition of the original high-dimensional problem to a number of lower-dimensional subproblems. Specifically, this criterion checks all the existing interactions between input parameters of a specific problem based on the high-dimensional model representation (HDMR) method, and therefore automatically provides the subproblems which only involve interacting input parameters. The efficiency of the approach is demonstrated using examples of both smooth and non-smooth problems with number of input parameters up to 500, and the approach is compared against other existing algorithms.

Finite Element Modeling of Fatigue in Fiber–Metal Laminates

Innovative hybrid materials developed at Delft University of Technology (e.g., ARALL and GLARE) dramatically reduce life-cycle costs and offer a great opportunity for service life extension of legacy aircraft. Replacement or repair of damaged aircraft components requires high-strength composite materials with high tailorability, fatigue, and impact-damage resistance, all of which are offered by the advanced hybrid materials. In addition, a reliable fatigue-life evaluation methodology for hybrid structures of arbitrary layup, configuration, constituent materials, and geometry is necessary. An efficient computational framework is presented for simulation of fatigue fracture in fiber–metal laminates based on the homogenized laminate modeled with large shell elements and cohesive zone used to simulate crack propagation. The cohesive traction–separation relationship is calibrated against the analytical solution for the strain-energy release rate, which explicitly accounts for the effect of fiber bridging. Appr...

Uncertainty quantification for complex systems with very high dimensional response using Grassmann manifold variations

Abstract This paper addresses uncertainty quantification (UQ) for problems where scalar (or low-dimensional vector) response quantities are insufficient and, instead, full-field (very high-dimensional) responses are of interest. To do so, an adaptive stochastic simulation-based methodology is introduced that refines the probability space based on Grassmann manifold variations. The proposed method has a multi-element character discretizing the probability space into simplex elements using a Delaunay triangulation. For every simplex, the high-dimensional solutions corresponding to its vertices (sample points) are projected onto the Grassmann manifold. The pairwise distances between these points are calculated using appropriately defined metrics and the elements with large total distance are sub-sampled and refined. As a result, regions of the probability space that produce significant changes in the full-field solution are accurately resolved. An added benefit is that an approximation of the solution within each element can be obtained by interpolation on the Grassmann manifold. The method is applied to study the probability of shear band formation in a bulk metallic glass using the shear transformation zone theory.

The effect of prior probabilities on quantification and propagation of imprecise probabilities resulting from small datasets

Abstract This paper outlines a methodology for Bayesian multimodel uncertainty quantification (UQ) and propagation and presents an investigation into the effect of prior probabilities on the resulting uncertainties. The UQ methodology is adapted from the information-theoretic method previously presented by the authors (Zhang and Shields, 2018) to a fully Bayesian construction that enables greater flexibility in quantifying uncertainty in probability model form. Being Bayesian in nature and rooted in UQ from small datasets, prior probabilities in both probability model form and model parameters are shown to have a significant impact on quantified uncertainties and, consequently, on the uncertainties propagated through a physics-based model. These effects are specifically investigated for a simplified plate buckling problem with uncertainties in material properties derived from a small number of experiments using noninformative priors and priors derived from past studies of varying appropriateness. It is illustrated that prior probabilities can have a significant impact on multimodel UQ for small datasets and inappropriate (but seemingly reasonable) priors may even have lingering effects that bias probabilities even for large datasets. When applied to uncertainty propagation, this may result in probability bounds on response quantities that do not include the true probabilities.

Reviewer's Recognition

The Editor and Editorial Board of the Journal of Risk and Uncertainty in Engineering Systems, Part B: Mechanical Engineering would like to thank all of the reviewers for volunteering their expertise and time reviewing manuscripts in 2017. Serving as reviewers for the journal is a critical service necessary to maintain the quality of our publication and to provide the authors with a valuable peer review of their work. Below is a complete list of reviewer’s for 2017. We would also like to acknowledge two outstanding Reviewers of the Year.