Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Stochastic Uncertainty Quantification”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

404 records · Page 23

Multiscale dynamics of reactive fronts in the subsurface

1. Brief summary of progress to date. Major Goals of the Project. Overreaching goals of the proposed activity are to develop an or- ganizing framework to describe nonlinear reactive transport in geologic systems over many length and time scales, and corresponding physics-based adaptive multi-scale hybrid models. We will also provide computational tools for robust uncertainty quantification in hybrid simulations. Major Activities of the Project. The major activities in this reporting period have focused on developing a time-averaging framework for ADR transport (Task 1) and global diagnosis criteria for spatio-temporal algorithm refinement (Task 2), as well as numerical validation and verification of the analytical approaches developed. The main activities are as follows: (a) We use homogenization theory to perform a spatio-temporal upscaling of pore-scale advection- diffusion equation in a periodic porous medium in the presence of fluctuating boundary conditions on some parts of the boundary and nonlinear chemical reaction on the solid-liquid interface and we investigate the constraints under which the macroscopic equation gives an accurate explanation of the underlying physics at the pore scale. We demonstrate that the dynamics at the continuum scale is strongly influenced by the interplay between signal frequency at the boundary and transport processes at the pore level. Furthermore, we show that by choosing appropriate scaling factors, a homogenized equation along with expressions for the effective coefficients reliably describe the concentration of solute at the length and time scales vastly different from physically robust micro-scale provided that some specific scale separation constraints are met. Our analysis proves that if the reaction rate or advection at the micro-scale are large, the effectiveness of the homogenized model is impeded. (b) We test numerically the conditions derived theoretically for advective-diffusive and reactive trans- port in a fracture with planar walls subject to temporally oscillating Dirichlet boundary conditions. The numerical analysis confirms the theoretical predictions. Significant Results. Two paper are in preparation. 2. Description of work yet to be accomplished The Tasks yet to be accomplished are listed below: (a) Quantify the impact of uncertainty in wall geometry of planar fractures on predictions hybrids of ARD transport (b) Develop coupling conditions for time hybridzation (c) Quantify the impact of uncertainty in pore geometry on predictions of hybrids of ARD transport in porous media, i.e. develop stochastic hybrid models (d) Use the hybrids combined with X-Ray tomographic images of pore scale structure (from Dr. Gouze, see letter of collaboration) to model dynamic changes in porosity/permeability

42 ENGINEERING↗

Jensen–Shannon divergence based novel loss functions for Bayesian neural networks

Bayesian neural networks (BNNs) are state-of-the-art machine learning methods that can naturally regularize and systematically quantify uncertainties using their stochastic parameters. Kullback–Leibler (KL) divergence-based variational inference used in BNNs suffer from unstable optimization and challenges in approximating light-tailed posteriors due to the unbounded nature of the KL divergence. To resolve these issues, we formulate a novel loss function for BNNs based on a new modification to the generalized Jensen–Shannon (JS) divergence, which is bounded. In addition, we propose a Geometric JS divergence-based loss, which is computationally efficient since it can be evaluated analytically. We found that the JS divergence-based variational inference is intractable, and hence employed a constrained optimization framework to formulate these losses. Our theoretical analysis and empirical experiments on multiple regression and classification data sets suggest that the proposed losses perform better than the KL divergence-based loss, especially when the data sets are noisy or biased. Specifically, there are approximately 5% and 8% improvements in accuracy for a noise-added CIFAR-10 dataset and a regression dataset, respectively. There is about 13% reduction in false negative predictions of a biased histopathology dataset. Additionally, we quantify and compare the uncertainty metrics for the regression and classification tasks.

97 MATHEMATICS AND COMPUTING↗

Differentiable Multiphysics Codes: A Breakthrough Technology for Simulation and Computing

This document summarizes the findings of a strategic planning exercise commissioned by the Weapons Simulation and Computing, Computational Physics (WSC/CP) program at the Lawrence Livermore National Laboratory (LLNL) in FY24. During the year, the committee met with multiple stakeholder communities to gather input, opinions, suggestions and concerns which have been incorporated throughout this document. The key findings from this exercise are summarized: • The development of multiphysics modelling and simulation (mod/sim) codes and software technologies, their deployment on exascale compute platforms, and their broad adoption across the NNSA is a major success of the Advanced Simulation and Computing (ASC) program and the Exascale Computing Project (ECP). Sustained investment in these core technologies is essential. • Today’s state of the art involves running ensembles of O(100K) simulations to perform uncertainty quantification (UQ) and design studies using multiple statistical methods such as Bayesian optimization to understand sensitivities of our models and explore parameterized design spaces. Even with exascale computing, we are practically limited to O(10) parameters in these studies since the number of simulations required to sample the space scales exponentially with the number of design parameters. • The data from these simulation ensembles is increasingly being used to train machine learned (ML) surrogates (or reduced order models, ROMs) which can then be used for optimization or real time design exploration. However, the trained surrogates are still limited in the number of parameters they can represent due to the sampling limitations previously noted. • Augmenting our suite of integrated multiphysics simulation codes, both current and emerging, with the ability to compute gradients (solution derivatives) of arbitrary simulation outputs with respect to (some or all) simulation inputs would be a breakthrough technology, opening the door to a new era of efficient and automated inverse design based on verified and validated mod/sim capabilities. • This capability, which we refer to as differentiable multiphysics codes (DMCs), would revolutionize both UQ and optimization studies by breaking the curse of dimensionality that presently limits our “gradient-free” ensemble based computing approach. A similar breakthrough occurred in the AI/ML community once the ability to compute gradients of arbitrary loss functions using back-propagation became commonplace. Gradient information from the multiphysics codes can also be used to dramatically improve the efficiency and scale of training of ML/ROM surrogates for rapid assessments. • Achieving this in our suite of codes will be a grand challenge, similar to the amount of effort that was required to transition from CPU to GPU computing. It will require buy-in from the entire WSC/CP program and beyond, including all integrated codes, physics and engineering models, third-party library dependencies and performance portability abstractions. It will also require investment in research and development of numerical methods for computing adjoints of coupled physics across multiple adaptively refined moving meshes and of stochastic (Monte Carlo) and mesh free (SPH) methods. • New software and numerical techniques, largely pioneered by the AI/ML community, make this feasible. Chief among these is automatic differentiation (AD), the ability to employ AD at point-wise locations in a physics calculation (instead of traditional black-box approaches) and the ability to perform “back-propagation in time” (or reverse mode AD) for non-linear partial differential equations (PDEs). Fundamentally, the conclusion of this strategic planning exercise is that the time is right to undertake a large scale effort in WSC, centered on the existing integrated codes, to continue the natural evolution of mod/sim in the age of AI/ML. Instead of attempting to replace mod/sim with purely data driven AI/ML models, we believe the key to success is to integrate AI/ML by building on top of the decades of hard-won knowledge and the verified/validated multiphysics modelling capability that is the hallmark of the ASC program.

97 MATHEMATICS AND COMPUTING↗

Parameter estimation with maximal updated densities

A recently developed measure-theoretic framework solves a stochastic inverse problem (SIP) for models where uncertainties in model output data are predominantly due to aleatoric (i.e., irreducible) uncertainties in model inputs (i.e., parameters). The subsequent inferential target is a distribution on parameters. Another type of inverse problem is to quantify uncertainties in estimates of “true” parameter values under the assumption that such uncertainties should be reduced as more data are incorporated into the problem, i.e., the uncertainty is considered epistemic. A major contribution of this work is the formulation and solution of such a parameter identification problem (PIP) within the measure-theoretic framework developed for the SIP. The approach is novel in that it utilizes a solution to a stochastic forward problem (SFP) to update an initial density only in the parameter directions informed by the model output data. In other words, this method performs “selective regularization” only in the parameter directions not informed by data. Further, the solution is defined by a maximal updated density (MUD) point where the updated density defines the measure-theoretic solution to the PIP. Another significant contribution of this work is the full theory of existence and uniqueness of MUD points for linear maps with Gaussian distributions. Data-constructed Quantity of Interest (QoI) maps are also presented and analyzed for solving the PIP within this measure-theoretic framework as a means of reducing uncertainties in the MUD estimate. We conclude with a demonstration of the general applicability of the method on two problems involving either spatial or temporal data for estimating uncertain model parameters. The first problem utilizes spatial data from a stationary partial differential equation to produce a MUD estimate of an uncertain boundary condition. The second problem utilizes temporal data obtained from the state-of-the-art ADvanced CIRCulation (ADCIRC) model to obtain a MUD estimate of uncertain wind drag coefficients for a simulated extreme weather event near the Shinnecock Inlet located in the Outer Barrier of Long Island, NY, USA.

42 ENGINEERING↗

Coupled land surface–subsurface hydrogeophysical inverse modeling to estimate soil organic carbon content and explore associated hydrological and thermal dynamics in the Arctic tundra

Quantitative characterization of soil organic carbon (OC) content is essential due to its significant impacts on surface–subsurface hydrological–thermal processes and microbial decomposition of OC, which both in turn are important for predicting carbon–climate feedbacks. While such quantification is particularly important in the vulnerable organic-rich Arctic region, it is challenging to achieve due to the general limitations of conventional core sampling and analysis methods, and to the extremely dynamic nature of hydrological–thermal processes associated with annual freeze–thaw events. In this study, we develop and test an inversion scheme that can flexibly use single or multiple datasets – including soil liquid water content, temperature and electrical resistivity tomography (ERT) data – to estimate the vertical distribution of OC content. Our approach relies on the fact that OC content strongly influences soil hydrological–thermal parameters and, therefore, indirectly controls the spatiotemporal dynamics of soil liquid water content, temperature and their correlated electrical resistivity. We employ the Community Land Model to simulate nonisothermal surface–subsurface hydrological dynamics from the bedrock to the top of canopy, with consideration of land surface processes (e.g., solar radiation balance, evapotranspiration, snow accumulation and melting) and ice–liquid water phase transitions. For inversion, we combine a deterministic and an adaptive Markov chain Monte Carlo (MCMC) optimization algorithm to estimate a posteriori distributions of desired model parameters. For hydrological–thermal-to-geophysical variable transformation, the simulated subsurface temperature, liquid water content and ice content are explicitly linked to soil electrical resistivity via petrophysical and geophysical models. We validate the developed scheme using different numerical experiments and evaluate the influence of measurement errors and benefit of joint inversion on the estimation of OC and other parameters. We also quantify the propagation of uncertainty from the estimated parameters to prediction of hydrological–thermal responses. We find that, compared to inversion of single dataset (temperature, liquid water content or apparent resistivity), joint inversion of these datasets significantly reduces parameter uncertainty. We find that the joint inversion approach is able to estimate OC and sand content within the shallow active layer (top 0.3 m of soil) with high reliability. Due to the small variations of temperature and moisture within the shallow permafrost (here at about 0.6 m depth), the approach is unable to estimate OC with confidence. However, if the soil porosity is functionally related to the OC and mineral content, which is often observed in organic-rich Arctic soil, the uncertainty of OC estimate at this depth remarkably decreases. Our study documents the value of the new surface–subsurface, deterministic–stochastic inversion approach, as well as the benefit of including multiple types of data to estimate OC and associated hydrological–thermal dynamics.

58 GEOSCIENCES↗

Validation and uncertainty of DRF's for a 1"×2" NaI collimated detector for radioisotope holdup measurements

Detector response functions (DRFs) have become the subject of increasing scientific interest for the last thirty years in several industrial applications of radiation detection. These applications include gamma-ray, prompt gamma-ray, and X-ray spectrometry for elemental analysis and location as applied to mining, radio-tracing in medicine, and holdup source characterization. However, it is difficult to create a rigorous mathematical formulation of the DRF. Therefore, semi-empirical and stochastic approaches have been developed instead to construct detector and application specific DRFs. A DRF can be considered to be a function that converts the energy-dependent incidence of incoming source particles onto a detector into a detector response spectrum that simulates the experimental response. The DRF can also be used in the reverse sense in an inverse problem setting, as a step in the process of predicting the physical characteristics of an unknown source (e.g. the holdup problem and library spectra). Much of the recent work on DRFs has been performed by Gardner. He has developed accurate DRF models through semi-empirical curve fitting and Monte Carlo simulation. His DRFs for 3" X 3" NaI detectors have been developed with the experimental measurements taken by Heath and for 6"x6" bare (unshielded) NaI(Tl) detector data with single-energy sources centered on the detector face at a distance of 10 cm. There was agreement with the Heath benchmark detector measurements within two Poisson standard deviations of the measured data. DRFs have also been developed for other detectors such as high purity Germanium detectors and Si(Li) for the same geometries as NaI. Other than a few artifacts such as backscatter peaks, there was good agreement between simulated responses and the experimental data.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

DiaMonD: An Integrated Multifaceted Approach to Mathematics at the Interfaces of Data, Models, and Decisions

For many problems in computational science and engineering, the so-called forward problem—solution of the underlying mathematical model to yield output quantities of interest, given input parameters—is difficult enough for frontier complex models, which are often characterized by coupled multiphysics and possibly stochastic behavior over a wide range of length and time scales. However, we face the need to move beyond the forward problem, to address what is often the ultimate goal of CS&E: decision making. This requires us to tackle a spectrum of mathematical problems that subsume, and are thus even more difficult than, the forward problem. First, given experimental data, we wish to estimate unknown parameters characterizing a model of a natural or engineered system by solving an inverse problem. Second, we seek the optimal configuration of the system (or experiments) by solving an optimal design problem. Third, the optimal operation of the system must be determined by solving an optimal control problem. And fourth, we must quantify uncertainties as they propagate through all of the preceding problems, from data to model inference to prediction and finally to optimal design and control. Optimal mathematical methods and algorithms for such end-to-end, data-to-decisions modeling and simulation of complex problems—requiring integrated solution of forward, inverse, optimization, and UQ problems for large, multiphysics, multiscale models—entails challenges of the highest order, while also presenting great opportunities for applied mathematics research. Unfortunately, research on forward, inverse, optimization, and UQ problems has all-too-often progressed in isolation. This has led to mathematical methods that perform well when considering the forward simulation in isolation, but are prohibitive or suboptimal or unstable when combined with other methods within the framework of inversion, optimization, or UQ. The reverse is also true: general-purpose methods developed within the inversion, optimization, and UQ fields often become prohibitive when applied to complex CS&E problems since they do not exploit their structure.

58 GEOSCIENCES↗

Advancing Multi-Hazard Risk and Safety Considerations for Aging Nuclear Facilities (Revision 1)

This project will demonstrate a multi-hazard time-dependent probabilistic risk assessment (PRA) approach for nuclear facilities considering aging-related deterioration of structures. A generic pressurized water reactor (PWR) reactor subjected to seismic mainshock-aftershock sequences considering the aging of the containment structure will be used as a case study to demonstrate the multi-hazard PRA approach. Using advanced modeling and simulation, seismic mainshock-aftershock fragility functions will be simulated for the containment structure considering aging effects. A multi-hazard PRA model for a generic PWR reactor will be built to quantify the multi-hazard core damage frequency (CDF) and large early release frequency (LERF) with explicit time-dependent modeling of event sequences. To date, the cascading impacts of multi-hazards are not adequately accounted for in the PRA models for nuclear facilities. In addition, for both initial and periodic evaluation of facilities to withstand natural phenomena hazards (NPH), deterioration of the SSCs due to aging and other effects is not adequately considered. By advancing multi-hazard considerations accounting for aging effects, this project will contribute to improved understanding of the safety of aging nuclear facilities. Project outcomes such as the multi-hazard CDF and LERF will also allow facility owners to optimize upgrade and retrofit protocols. This project is divided into two components: (1) advanced modeling and simulations; and (2) multi-hazard PRA. For the first component, Idaho National Laboratory’s (INL) Multi-hazard Analysis for STOchastic time-DOmaiN phenomena (MASTODON) and BlackBear codes will be used to simulate the seismic response and damage of a containment structure under cascading mainshock-aftershock sequences with aging effects. Uncertainties related to the seismic inputs, material parameters, and environmental factors such as temperature and humidity will be identified and propagated to the fragility functions. These fragility functions, which consider aging effects, will be time dependent. The second component will take the fragility information from the first component to build a multi-hazard time-dependent PRA model starting from a generic PWR model to evaluate the multi-hazard CDF and LERF and characterize the associated consequences. For this component, OpenPRA Web Application and SAPHIRE code will be used. Events such as the Fukushima Daiichi accident have highlighted the importance of considering the cascading impacts of multi-hazards for PRA. Moreover, many of the reactors in the current nuclear fleet in the United States (US) are already operating well beyond their initially planned design life, and applications for further extensions to operating licenses are being considered. Therefore, considering multi-hazard effects and aging deterioration in the NPH risk assessment process will contribute to the safety of both existing and future nuclear facilities. Outcomes of this project will thus directly benefit the standards DOE-STD-1020-16 and DOE-HDBK-1224-18.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗