Stochastic collocation (SC) is a well-known non-intrusive method of constructing surrogate models for uncertainty quantification. In dynamical systems, SC is especially suited for full-field uncertainty propagation that characterizes the distributions of the high-dimensional solution fields of a model with stochastic input parameters. However, due to the highly nonlinear nature of the parameter-to-solution map in even the simplest dynamical systems, the constructed SC surrogates are often inaccurate. This work presents an alternative approach, where we apply the SC approximation over the dynamics of the model, rather than the solution. By combining the data-driven sparse identification of nonlinear dynamics framework with SC, we construct dynamics surrogates and integrate them through time to construct the surrogate solutions. We demonstrate that the SC-over-dynamics framework leads to smaller errors, both in terms of the approximated system trajectories as well as the model state distributions, when compared against full-field SC applied to the solutions directly. We present numerical evidence of this improvement using three test problems: a chaotic ordinary differential equation, and two partial differential equations from solid mechanics.
Composite materials with different microstructural material symmetries are common in engineering applications where grain structure, alloying and particle/fiber packing are optimized via controlled manufacturing. In fact these microstructural tunings can be done throughout a part to achieve functional gradation and optimization at a structural level. To predict the performance of particular microstructural configuration and thereby overall performance, constitutive models of materials with microstructure are needed. In this work we provide neural network architectures that provide effective homogenization models of materials with anisotropic components. These models satisfy equivariance and material symmetry principles inherently through a combination of equivariant and tensor basis operations. We demonstrate them on datasets of stochastic volume elements with different textures and phases where the material undergoes elastic and plastic deformation, and show that the these network architectures provide significant performance improvements.
A combined Mode I-II cohesive zone (CZ) elasto-plastic constitutive model, and a two-dimensional (2D) cohesive interface element (CIE) are formulated and implemented at small strain within an ABAQUS User Element (UEL) for simulating 2D crack nucleation and propagation in fluid-saturated porous media. The CZ model mitigates problems of convergence for the global Newton-Raphson solver within ABAQUS, which when combined with a viscous stabilization procedure allows for simulation of post-peak response under load control for coupled poromechanical finite element analysis, such as concrete gravity dam stability analysis. Verification examples are presented, along with a more complex ambient limestone-concrete wedge fracture experiment, water-pressurized concrete wedge experiment, and concrete gravity dam stability analyses. A calibration procedure for estimating the CZ parameters is demonstrated with the limestone-concrete wedge fracture process. For the water-pressurized concrete wedge fracture experiment it is shown that the inherent time-dependence of the poromechanical CIE analysis provides a good match with experimental force versus displacement results at various crack mouth opening rates, yet misses the pore water pressure evolution ahead of the crack tip propagation. This is likely a result of the concrete being partially-saturated in the experiment, whereas the finite element analysis assumes fully water saturated concrete. For the concrete gravity dam analysis, it is shown that base crack opening and associated water uplift pressure leads to a reduced Factor of Safety, which is confirmed by separate analytical calculations.
Machine-learning function representations such as neural networks have proven to be excellent constructs for constitutive modeling due to their flexibility to represent highly nonlinear data and their ability to incorporate constitutive constraints, which also allows them to generalize well to unseen data. In this work, we extend a polyconvex hyperelastic neural network framework to (isotropic) thermo-hyperelasticity by specifying the thermodynamic and material theoretic requirements for an expansion of the Helmholtz free energy expressed in terms of deformation invariants and temperature. Different formulations which a priori ensure polyconvexity with respect to deformation and concavity with respect to temperature are proposed and discussed. The physics-augmented neural networks are furthermore calibrated with a recently proposed sparsification algorithm that not only aims to fit the training data but also penalizes the number of active parameters, which prevents overfitting in the low data regime and promotes generalization. The performance of the proposed framework is demonstrated on synthetic data, which illustrate the expected thermomechanical phenomena, and existing temperature-dependent uniaxial tension and tension-torsion experimental datasets.
During fracture amorphous oxides exhibit irreversible processes, including inelastic and nonrecoverable relaxation effects in the process zone surrounding the crack tip. Here, classical molecular dynamics simulations were used with a reactive forcefield to evaluate inelastic relaxation processes in five amorphous sodium silicate compositions. Overall, the 20% Na2O-SiO2(NS20) composition exhibited the most inelastic relaxation, followed by the 15% Na2O-SiO2(NS15) composition, the 25% Na2O-SiO2(NS25) composition, and finally the 10% (NS10) and 30% (NS30) Na2O-SiO2 compositions. Coordination analysis of the Na+ ions identified that during inelastic relaxation the Na+ ions were increasingly coordinated by nonbridging oxygens (NBOs) for the NS10 and NS15 compositions, which was supported by radial analysis of the O-Na-O bond angles surrounding the crack tip. Across the sodium silicate compositional range, two different inelastic relaxation mechanism were identified based on the amount of bridging oxygens (BOs) and NBOs in the Na+ ion coordination shell. At lower (NS10) and higher (NS30) sodium compositions, the entire structured relaxed toward the crack tip. In contrast at intermediate sodium concentrations (NS20) the Na+ ion migrates toward the crack tip separately from the network structure. By developing a fundamental understanding of how modified silica systems respond to static stress fields, we will be able to predict how varying amorphous silicate systems exhibit slow crack growth.
Contact mechanics, or the modeling of the impenetrability of solid objects, is fundamental to computational solid mechanics (CSM) applications yet is oftentimes the most challenging in terms of computational efficiency and performance. These challenges arise from the irregularity and highly dynamic nature of contact simulation, particularly with algorithms designed for distributed memory architectures. First among these challenges is the inherent load imbalance when distributing contact load across compute nodes. This imbalance is highly problem dependent, and relates to the surface area of contact manifolds and the volume around them, rather than the distribution of the mesh over compute nodes, meaning the application load can vary drastically over different phases. The dynamic nature of contact problems motivates the use of distributed asynchronous many-tasking (AMT) frameworks to efficiently handle irregular workloads. In this paper, we present our work on distBVH, a distributed contact solution using the DARMA/vt library for asynchronous tasking that is also capable of running on-node Kokkos-based kernels. We explore how distBVH addresses the various challenges of CSM contact problems. We evaluate the use of many of DARMA/vt’s dynamic load balancers and demonstrate how our load balancing approach can provide significant performance improvements on various computational solid mechanics benchmarks. Additionally, we show how our approach can take advantage of DARMA/vt for tasking and efficient on-node kernels using Kokkos to scale over hundreds of processing elements.
There is an increasing aspiration to utilize machine learning (ML) for various tasks of relevance to national security. ML models have thus far been mostly applied to tasks and domains that, while impactful, have sufficient volume of data. For predictive tasks of national security relevance, ML models of great capacity (ability to approximate nonlinear trends in input-output maps) are often needed to capture the complex underlying physics. However, scientific problems of relevance to national security are often accompanied by various sources of sparse and/or incomplete data, including experiments and simulations, across different regimes of operation, of varying degrees of fidelity, and include noise with different characteristics and/or intensity. State-of-the-art ML models, despite exhibiting superior performance on the task and domain they were trained on, may suffer detrimental loss in performance in such sparse data environments. This report summarizes the results of the Laboratory Directed Research and Development project entitled Trust-Enhancing Probabilistic Transfer Learning for Sparse and Noisy Data Environments. The objective of the project was to develop a new transfer learning (TL) framework that aims to adaptively blend the data across different sources in tackling one task of interest, resulting in enhanced trustworthiness of ML models for mission- and safety-critical systems. The proposed framework determines when it is worth applying TL and how much knowledge is to be transferred, despite uncontrollable uncertainties. The framework accomplishes this by leveraging concepts and techniques from the fields of Bayesian inverse modeling and uncertainty quantification, relying on strong mathematical foundations of probability and measure theories to devise new uncertainty-aware TL workflows.
Predictive modeling typically relies on Bayesian model calibration to provide uncertainty quantification. Variational inference utilizing fully independent (“mean-field”) Gaussian distributions are often used as approximate probability density functions. This simplification is attractive since the number of variational parameters grows only linearly with the number of unknown model parameters. However, the resulting diagonal covariance structure and unimodal behavior can be too restrictive to provide useful approximations of intractable Bayesian posteriors that exhibit highly non-Gaussian behavior, including multimodality. High-fidelity surrogate posteriors for these problems can be obtained by considering the family of Gaussian mixtures. Gaussian mixtures are capable of capturing multiple modes and approximating any distribution to an arbitrary degree of accuracy, while maintaining some analytical tractability. Unfortunately, variational inference using Gaussian mixtures with full-covariance structures suffers from a quadratic growth in variational parameters with the number of model parameters. The existence of multiple local minima due to strong nonconvex trends in the loss functions often associated with variational inference present additional complications, These challenges motivate the need for robust initialization procedures to improve the performance and computational scalability of variational inference with mixture models. In this work, we propose a method for constructing an initial Gaussian mixture model approximation that can be used to warm-start the iterative solvers for variational inference. The procedure begins with a global optimization stage in model parameter space. In this step, local gradient-based optimization, globalized through multistart, is used to determine a set of local maxima, which we take to approximate the mixture component centers. Around each mode, a local Gaussian approximation is constructed via the Laplace approximation. Finally, the mixture weights are determined through constrained least squares regression. The robustness and scalability of the proposed methodology is demonstrated through application to an ensemble of synthetic tests using high-dimensional, multimodal probability density functions. Here, the practical aspects of the approach are demonstrated with inversion problems in structural dynamics.
The development of highly accurate constitutive models for materials that undergo path-dependent processes continues to be a complex challenge in computational solid mechanics. Challenges arise both in considering the appropriate model assumptions and from the viewpoint of data availability, verification, and validation. Recently, data-driven modeling approaches have been proposed that aim to establish stress-evolution laws that avoid user-chosen functional forms by relying on machine learning representations and algorithms. However, these approaches not only require a significant amount of data but also need data that probes the full stress space with a variety of complex loading paths. Furthermore, they rarely enforce all necessary thermodynamic principles as hard constraints. Hence, they are in particular not suitable for low-data or limited-data regimes, where the first arises from the cost of obtaining the data and the latter from the experimental limitations of obtaining labeled data, which is commonly the case in engineering applications. In this work, we discuss a hybrid framework that can work on a variable amount of data by relying on the modularity of the elastoplasticity formulation where each component of the model can be chosen to be either a classical phenomenological or a data-driven model depending on the amount of available information and the complexity of the response. The method is tested on synthetic uniaxial data coming from simulations as well as cyclic experimental data for structural materials. The discovered material models are found to not only interpolate well but also allow for accurate extrapolation in a thermodynamically consistent manner far outside the domain of the training data. This ability to extrapolate from limited data was the main reason for the early and continued success of phenomenological models and the main shortcoming in machine learning-enabled constitutive modeling approaches. Training aspects and details of the implementation of these models into Finite Element simulations are discussed and analyzed.
Reactive classical molecular dynamics simulations of sodium silicate glasses, xNa2O–(100 − x)SiO2 (x = 10–30), under quasi-static loading, were performed for the analysis of molecular scale fracture mechanisms. Mechanical properties of the sodium silicate glasses were consistent with experimentally reported values, and the amount of crack propagation varied with reported fracture toughness values. The most crack propagation occurred in NS20 systems (20-mol% Na2O) compared with the other simulated compositions. Dissipation via two mechanisms, the first through sodium migration as a lower activation energy process and the second through structural rearrangement as a higher activation energy process, was calculated and accounted for the energy that was not stored elastically or associated with the formation of new fracture surfaces. A correlation between crack propagation and energy dissipation was identified, with systems with higher crack propagation exhibiting less energy dissipation. Sodium silicate glass compositions with lower energy dissipation also exhibited the most sodium movement and structural rearrangement within 10 Å of the crack tip during loading. Therefore, high sodium mobility near the crack tip may enable energy dissipation without requiring formation of structural defects. Therefore, the varying mobilities of the network modifiers near crack tips influence the brittleness and the crack growth rate of modified amorphous oxide systems.
Rock, concrete, and other engineered materials are often composed of several minerals that change volumetrically in response to variations in the moisture content of the local environment. Such differential shrinkage is caused by varying shrinkage rates between mineral compositions during dehydration. Using both 3D X-ray imaging of geo-architected samples and peridynamic (PD) numerical simulations, we show that the spatial distribution of the clay affects the crack network geometry with distributed clay particles yielding the most complex crack networks and percent damage (99.56%), along with a 60% reduction in material strength. We also demonstrate that crack formation, growth, coalescence, and distribution during dehydration, are controlled by the differential shrinkage rates between a highly shrinkable clay and a homogeneous mortar matrix. Sensitivity tests performed with the PD models show a clay shrinkage parameter of 0.4 yields considerable damage, and reductions in the parameter can result in a significant reduction in fracturing and an increase in material strength. Additionally, isolated clay inclusions induced localized fracturing predominantly due to debonding between the clay and matrix. These insights indicate differential shrinkage is a source of potential failure in natural and engineered barriers used to sequester anthropogenic waste.
This report details a new method for propagating parameter uncertainty (forward uncertainty quantification) in partial differential equations (PDE) based computational mechanics applications. The method provides full-field quantities of interest by solving for the joint probability density function (PDF) equations which are implied by the PDEs with uncertain parameters. Full-field uncertainty quantification enables the design of complex systems where quantities of interest, such as failure points, are not known apriori. The method, motivated by the well-known probability density function (PDF) propagation method of turbulence modeling, uses an ensemble of solutions to provide the joint PDF of desired quantities at every point in the domain. A small subset of the ensemble is computed exactly, and the remainder of the samples are computed with approximation of the driving (dynamics) term of the PDEs based on those exact solutions. Although the proposed method has commonalities with traditional interpolatory stochastic collocation methods applied directly to quantities of interest, it is distinct and exploits the parameter dependence and smoothness of the dynamics term of the governing PDEs. The efficacy of the method is demonstrated by applying it to two target problems: solid mechanics explicit dynamics with uncertain material model parameters, and reacting hypersonic fluid mechanics with uncertain chemical kinetic rate parameters. A minimally invasive implementation of the method for representative codes SPARC (reacting hypersonics) and NimbleSM (finite- element solid mechanics) and associated software details are described. For solid mechanics demonstration problems the method shows order of magnitudes improvement in accuracy over traditional stochastic collocation. For the reacting hypersonics problem, the method is implemented as a streamline integration and results show very good accuracy for the approximate sample solutions of re-entry flow past the Apollo capsule geometry at Mach 30.
Brittle material failure in high consequence systems can appear random and unpredictable at subcritical stresses. Gaps in our understanding of how structural flaws and environmental factors (humidity, temperature) impact fracture propagation need to be addressed to circumvent this issue. A combined experimental and computational approach composed of molecular dynamics (MD) simulations, numerical modeling, and atomic force microscopy (AFM) has been undertaken to identify mechanisms of slow crack growth in silicate glasses. AFM characterization of crack growth as slow as 10-13 m/s was observed, with some stepwise crack growth. MD simulations have identified the critical role of inelastic relaxation in crack propagation, including evolution of the structure during relaxation. A numerical model for the existence of a stress intensity threshold, a stress intensity below which a fracture will not propagate, was developed. This transferrable model for predicting slow crack growth is being incorporated into mission-based programs.