Search NASA⌕ Search

SEARCH · Search NASA

Results for “model form 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.

At least 199 records · Page 11

Quantified uncertainty in thermodynamic modeling for materials design

Phase fractions, compositions and energies of the stable phases as a function of macroscopic composition, temperature, and pressure (X-T-P) are the principle correlations needed for the design of new materials and improvement of existing materials. They are the outcomes of thermodynamic modeling based on the CALculation of PHAse Diagrams (CALPHAD) approach. Here, the accuracy of CALPHAD predictions vary widely in X-T-P space due to experimental error, model inadequacy and unequal data coverage. In response, researchers have developed frameworks to quantify the uncertainty of thermodynamic property model parameters and propagate it to phase diagram predictions. In most previous studies, uncertainty was represented as intervals on phase boundaries (with respect to composition or temperature) and was unable to represent the uncertainty in invariant reactions or in the stability of phase regions. In this work, we propose a suite of tools that leverages samples from the multivariate model parameter distribution to represent uncertainty in forms that surpass previous limitations and are well suited to materials design. These representations include the distribution of phase diagrams and their features, as well as the dependence of phase stability and the distributions of phase fraction, composition, activity and Gibbs energy on X-T-P location - irrespective of the total number of components. Most critically, the new methodology allows the material designer to interrogate a certain composition and temperature domain and get in return the probability of different phases to be stable, which can positively impact materials design.

36 MATERIALS SCIENCE↗

Damage Assessment Technologies for Prognostics and Proactive Management of Materials Degradation (PMMD)

There are approximately 440 operating reactors in the global nuclear power plant (NPP) fleet with an average age greater than 20 years and design lives of 30 or 40 years. The United States is currently implementing license extensions of 20 years on many plants, and consideration is now being given to the concept of "life-beyond-60", license extension from 60 to 80 years and potentially longer. In almost all countries with NPPs, authorities are looking at some form of license renewal program. In support of NPP license renewal over the past decade, various national and international programs have been initiated. This paper discusses stressor-based prognostics and its role as part of emerging trends in Proactive Management of Materials Degradation (PMMD) applied to nuclear power plant structures, systems and components (SSC). The paper concisely explains the US Nuclear Regulatory Commission’s (NRC) program in PMMD, the basic principles of PMMD and its relationship to advanced diagnostics and prognostics. It then provides an assessment of the state of maturity for diagnostic and prognostic technologies, including NDE and related technologies for damage assessment, and the current trend to move from condition-based maintenance to on-line monitoring for advanced diagnostics and stressor-based prognostics. This development in technology requires advances in sensors; better understanding of what and how to measure within a nuclear power plant; enhanced data interrogation, communication and integration; new prediction models for damage/aging evolution; system integration for real-world deployments and quantification of uncertainties in what are inherently ill-posed problems. Stressor-based analysis is based upon understanding which stressor characteristics (e.g., pressure transients) provide a percussive indication that can be used for mapping subsequent damage due to a specific degradation mechanism. The resulting physical damage and the associated decrease in asset performance start with the application of a stressor to the component. The design engineer sets the desired operational stressor intensity level so that the degradation in the physical state of the component occurs slowly enough for the equipment to last for its required design life. In general, when the design limit of a stressor is exceeded (during operation), the component life expectancy starts to shorten. Conversely, careful control of operational parameters can enable extension of component life beyond that normally expected. For systems which were conservatively designed (such as nuclear power plants), the premise of the prognostic methodology is that a relationship can be derived that will allow a much more accurate projection of the remaining useful life. This is achieved by focusing on trending the stressor characterics rather than trending a performance metric. In this trend analysis example, the slope of the trended parameter is thought to give a measure of the degradation rate of the component performance. This is assumed to be a function of the rate of decline in the physical integrity of the equipment. Experience from measurements has shown this assumption to be true if one accounts for the nonlinearity which can occur between physical attributes and their effects on performance.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗

Elucidating the photodissociation fingerprint and quantifying the determination of organic hydroperoxides in gas-phase autoxidation

Hydroperoxides are formed in the atmospheric oxidation of volatile organic compounds, in the combustion autoxidation of fuel, in the cold environment of the interstellar medium, and also in some catalytic reactions. They play crucial roles in the formation and aging of secondary organic aerosols and in fuel autoignition. However, the concentration of organic hydroperoxides is seldom measured, and typical estimates have large uncertainties. In this work, we developed a mild and environmental-friendly method for the synthesis of alkyl hydroperoxides (ROOH) with various structures, and we systematically measured the absolute photoionization cross-sections (PICSs) of the ROOHs using synchrotron vacuum ultraviolet-photoionization mass spectrometry (SVUV-PIMS). A chemical titration method was combined with an SVUV-PIMS measurement to obtain the PICS of 4-hydroperoxy-2-pentanone, a typical molecule for combustion and atmospheric autoxidation ketohydroperoxides (KHPs). We found that organic hydroperoxide cations are largely dissociated by loss of OOH. This fingerprint was used for the identification and accurate quantification of the organic peroxides, and it can therefore be used to improve models for autoxidation chemistry. The synthesis method and photoionization dataset for organic hydroperoxides are useful for studying the chemistry of hydroperoxides and the reaction kinetics of the hydroperoxy radicals and for developing and evaluating kinetic models for the atmospheric autoxidation and combustion autoxidation of the organic compounds.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Quantifying Uncertainties of Microscopic Nuclear Theories

The goal of this project was to produce modernized nuclear data libraries of light ion fusion reactions and actinide fission product yields. Such libraries are essential input in both stockpile science but also fundamental science (nucleosynthesis, superheavy nuclei). When such libraries are used to model complex processes involving exotic, very short-lived nuclei that have not been measured experimentally, the underlying nuclear data come from theoretical models, the predictive power of which needs to be quantified, together with its uncertainties. By teaming up nuclear theorists and statisticians, we built comprehensive theoretical frameworks based on the most fundamental theories of fusion and fission that we coupled for the first time with a Bayesian framework with machine learning techniques for uncertainty quantification. Our results include: (1) statistical emulators for neutron-alpha scattering, and the fission product mass yields of neutron-induced fission of 239 Pu, which will enable developers to generate random realizations of the corresponding data for uncertainty quantification (UQ) in applications; (2) the first complete and most accurate ab initio prediction of the S-factor for proton capture on 7 Be, which – combined with UQ – may become the recommended value at solar energies in the next review of solar fusion cross sections; (3) statistical emulators for fission mass yields in the neutron-induced fission of 239 Pu; (4) the first quantification of how uncertainties in nuclear forces impact fission properties that are relevant in nucleosynthesis simulations to explain how heavy elements are formed in the Cosmos.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

CEMENTITIOUS BARRIERS PARTNERSHIP ACCOMPLISHMENTS AND RELEVANCE TO THE DOE COMPLEX

The Cementitious Barriers Partnership (CBP) was initiated to reduce risk and uncertainties in the performance assessments that directly impact U.S. Department of Energy (DOE) environmental cleanup and closure programs. The CBP is supported by the DOE Office of Environmental Management (DOE-EM) and has been specifically addressing the following critical EM program needs: (i) the long-term performance of cementitious barriers and materials in nuclear waste disposal facilities and (ii) increased understanding of contaminant transport behavior within cementitious barrier systems to support the development and deployment of adequate closure technologies. To accomplish this, the CBP has two initiatives: (1) an experimental initiative to increase understanding of changes in cementitious materials over long times (> 1000 years) over changing conditions and (2) a modeling initiative to enhance and integrate a set of computational tools validated by laboratory and field experimental data to improve understanding and prediction of the long-term performance of cementitious barriers and waste forms used in nuclear applications. In FY10, the CBP developed the initial phase of an integrated modeling tool that would serve as a screening tool which could help in making decisions concerning disposal and tank closure. The CBP experimental programs are underway to validate this tool and provide increased understanding of how CM changes over time and under changing conditions. These initial CBP products that will eventually be enhanced are anticipated to reduce the uncertainties of current methodologies for assessing cementitious barrier performance and increase the consistency and transparency of the DOE assessment process. These tools have application to low activity waste forms, high level waste tank closure, D&D and entombment of major nuclear facilities, landfill waste acceptance criteria, and in-situ grouting and immobilization of vadose zone contamination. This paper summarizes the recent work provided by the CBP to support DOE operations and regulatory compliance and the accomplishments over the past 2 years. Impacts of this work include: (1) a forum for DOE-NRC technical exchange, (2) material characterization to support PA predictions, (3) reducing uncertainty in PA predictions, (4) establishing base case performance to improve PA predictions, and (5) improving understanding and quantification of moisture and contaminant transport used in PAs. Additional CBP accomplishments include: sponsorship of a national test bed workshop to obtain collaboration in establishing the path forward in obtaining actual data to support future predictions on cementitious barrier performance evaluations, and participation in an International Atomic Energy Agency (IAEA) Cooperative Research Project on the use of cementitious barriers for low-level radioactive waste treatment and disposal.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗

Framework for Estimating Performance and Associated Uncertainty for Modified Aircraft Configurations

Flight testing has been the historical standard for determining aircraft airworthiness - however, increases in the cost of flight testing and the accuracy of inexpensive CFD promote certification by analysis to reduce or replace flight testing. A framework is introduced to predict the performance in the special case of a modification to an existing, previously certified aircraft. This framework uses a combination of existing flight test or high fidelity data of the original aircraft as well as lower fidelity data of the original and modified configurations. Two methods are presented which estimate the model form uncertainty of the modified configuration, which is then used to conduct non-deterministic simulations. The framework is applied to an example aircraft system with simulated flight test data to demonstrate the ability to predict the performance and associated uncertainty of modified aircraft configurations. However, it is important that the models and methods used are applicable and accurate throughout the intended use domain. The factors and limitations of the framework are explored to determine the range of applicability of the framework. The effects of these factors on the performance and uncertainty results are demonstrated using the example aircraft system. The framework is then applied to NASA’s X-57 Maxwell and each of its modifications. The estimated performance and associated uncertainties are then compared to the airworthiness criteria to evaluate the potential of the framework as a component to the certification by analysis process.

Uncertainty Quantification↗

Improvement in precision, accuracy, and efficiency in sstandardizing the characterization of granular materials

Useful prediction of the kinematics, dynamics, and chemistry of a system relies on precision and accuracy in the quantification of component properties, operating mechanisms, and collected data. In an attempt to emphasize, rather than gloss over, the benefit of proper characterization to fundamental investigations of multiphase systems incorporating solid particles, a set of procedures were developed and implemented for the purpose of providing a revised methodology having the desirable attributes of reduced uncertainty, expanded relevance and detail, and higher throughput. Better, faster, cheaper characterization of multiphase systems result. Methodologies are presented to characterize particle size, shape, size distribution, density (particle, skeletal and bulk), minimum fluidization velocity, void fraction, particle porosity, and assignment within the Geldart Classification. A novel form of the Ergun equation was used to determine the bulk void fractions and particle density. Accuracy of properties-characterization methodology was validated on materials of known properties prior to testing materials of unknown properties. Several of the standard present-day techniques were scrutinized and improved upon where appropriate. Validity, accuracy, and repeatability were assessed for the procedures presented and deemed higher than present-day techniques. A database of over seventy materials has been developed to assist in model validation efforts and future desig

42 ENGINEERING↗

Dose Uncertainty Quantification in AGC Experiment

This memorandum formally documents completion of the Advanced Reactor Technologies Level 3 Milestone (M3NT-18IN060504037), “Interoffice Memorandum detailing status of the AGC irradiation uncertainty analysis” due SeptemberAugust 30, 2018. This memorandum details the progress made toward the issuance of a final report detailing the AGC uncertainty analysis including the proposed methodology, progress on activities, and explanation of how the data will be used. Radiation damage estimation is an important component of the post-irradiation analysis of the AGC experiments. It depends primarily on the fast fluence, which is determined using well-established methods of spectral adjustments. These are based on best estimates from models (e.g. MCNP). Input cross-sections, and measured activities from flux wires in the experiment, propagate uncertainties in each of these. While the methods of propagating uncertainty are well-established, the final uncertainty estimates they provide are only as good as the estimates of uncertainty in the inputs on which they are based. The purpose of this work is to implement a method that eliminates some deficiencies in the ways input uncertainties are presently estimated. The rRadiation damage estimates for the AGC experiments are ostensibly based on experimental data in the form of flux wire activation measurements; each AGC experiment has been instrumented with flux wires at various locations along the axial length to provide this data. However, the estimatesy also depend on the results of MCNP calculations. To understand why, consider that the objective of the analysis is to estimate some number of parameters (the energy group fluences ?_i) that best reproduce the observations (the measured activities of the flux wires a_i).

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗

Bayesian learning with Gaussian processes for low-dimensional representations of time-dependent nonlinear systems

This work presents a data-driven method for learning low-dimensional time-dependent physics-based surrogate models whose predictions are endowed with uncertainty estimates. We use the operator inference approach to model reduction that poses the problem of learning low-dimensional model terms as a regression of state space data and corresponding time derivatives by minimizing the residual of reduced system equations. Standard operator inference models perform well with accurate training data that are dense in time, but producing stable and accurate models when the state data are noisy and/or sparse in time remains a challenge. Another challenge is the lack of uncertainty estimation for the predictions from the operator inference models. Our approach addresses these challenges by incorporating Gaussian process surrogates into the operator inference framework to (1) probabilistically describe uncertainties in the state predictions and (2) procure analytical time derivative estimates with quantified uncertainties. The formulation leads to a generalized least-squares regression and, ultimately, reduced-order models that are described probabilistically with a closed-form expression for the posterior distribution of the operators. The resulting probabilistic surrogate model propagates uncertainties from the observed state data to reduced-order predictions. Furthermore, we demonstrate the method is effective for constructing low-dimensional models of two nonlinear partial differential equations representing a compressible flow and a nonlinear diffusion–reaction process, as well as for estimating the parameters of a low-dimensional system of nonlinear ordinary differential equations representing compartmental models in epidemiology.

Data-driven model reduction↗

Development and Validation of a Lifecycle-based Prognostics Architecture with Test Bed Validation

On-line monitoring and tracking of nuclear plant system and component degradation is being investigated as a method for improving the safety, reliability, and maintainability of aging nuclear power plants. Accurate prediction of the current degradation state of system components and structures is important for accurate estimates of their remaining useful life (RUL). The correct quantification and propagation of both the measurement uncertainty and model uncertainty is necessary for quantifying the uncertainty of the RUL prediction. This research project developed and validated methods to perform RUL estimation throughout the lifecycle of plant components. Prognostic methods should seamlessly operate from beginning of component life (BOL) to end of component life (EOL). We term this "Lifecycle Prognostics." When a component is put into use, the only information available may be past failure times of similar components used in similar conditions, and the predicted failure distribution can be estimated with reliability methods such as Weibull Analysis (Type I Prognostics). As the component operates, it begins to degrade and consume its available life. This life consumption may be a function of system stresses, and the failure distribution should be updated to account for the system operational stress levels (Type II Prognostics). When degradation becomes apparent, this information can be used to again improve the RUL estimate (Type III Prognostics). This research focused on developing prognostics algorithms for the three types of prognostics, developing uncertainty quantification methods for each of the algorithms, and, most importantly, developing a framework using Bayesian methods to transition between prognostic model types and update failure distribution estimates as new information becomes available. The developed methods were then validated on a range of accelerated degradation test beds. The ultimate goal of prognostics is to provide an accurate assessment for RUL predictions, with as little uncertainty as possible. From a reliability and maintenance standpoint, there would be improved safety by avoiding all failures. Calculated risk would decrease, saving money by avoiding unnecessary maintenance. One major bottleneck for data-driven prognostics is the availability of run-to-failure degradation data. Without enough degradation data leading to failure, prognostic models can yield RUL distributions with large uncertainty or mathematically unsound predictions. To address these issues a "Lifecycle Prognostics" method was developed to create RUL distributions from Beginning of Life (BOL) to End of Life (EOL). This employs established Type I, II, and III prognostic methods, and Bayesian transitioning between each Type. Bayesian methods, as opposed to classical frequency statistics, show how an expected value, a priori, changes with new data to form a posterior distribution. For example, when you purchase a component you have a prior belief, or estimation, of how long it will operate before failing. As you operate it, you may collect information related to its condition that will allow you to update your estimated failure time. Bayesian methods are best used when limited data are available. The use of a prior also means that information is conserved when new data are available. The weightings of the prior belief and information contained in the sampled data are dependent on the variance (uncertainty) of the prior, the variance (uncertainty) of the data, and the amount of measured data (number of samples). If the variance of the prior is small compared to the uncertainty of the data, the prior will be weighed more heavily. However, as more data are collected, the data will be weighted more heavily and will eventually swamp out the prior in calculating the posterior distribution of model parameters. Fundamentally Bayesian analysis updates a prior belief with new data to get a posterior belief. The general approach to applying the Bayesian method to lifecycle prognostics consisted of identifying the prior, which is the RUL estimate and uncertainty from the previous prognostics type, and combining it with observational data related to the newer prognostics type. The resulting lifecycle prognostics algorithm uses all available information throughout the component lifecycle.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Improved multifidelity Monte Carlo estimators based on normalizing flows and dimensionality reduction techniques

Here, we study the problem of multifidelity uncertainty propagation for computationally expensive models. In particular, we consider the general setting where the high-fidelity and low-fidelity models have a dissimilar parameterization both in terms of number of random inputs and their probability distributions, which can be either known in closed form or provided through samples. We derive novel multifidelity Monte Carlo estimators which rely on a shared subspace between the high-fidelity and low-fidelity models where the parameters follow the same probability distribution, i.e., a standard Gaussian. We build the shared space employing normalizing flows to map different probability distributions into a common one, together with linear and nonlinear dimensionality reduction techniques, active subspaces and autoencoders, respectively, which capture the subspaces where the models vary the most. We then compose the existing low-fidelity model with these transformations and construct modified models with an increased correlation with the high-fidelity model, which therefore yield multifidelity estimators with reduced variance. A series of numerical experiments illustrate the properties and advantages of our approaches.

97 MATHEMATICS AND COMPUTING↗

Uncertainty in master-equation-based rate coefficient calculations

Rate coefficients are key quantities in gas phase kinetics and can be determined theoretically via master equation (ME) calculations. Rate coefficients characterize how fast a certain chemical species reacts away due to collisions into a specific product. Some of these collisions are simply transferring energy between the colliding partners, in which case the initial chemical species can undergo a unimolecular reaction: dissociation or isomerization. Other collisions are reactive, and the colliding partners either exchange atoms, these are direct reactions, or form complexes that can themselves react further or get stabilized by deactivating collisions with a bath gas. The input of MEs are molecular parameters: geometries, energies, and frequencies determined from ab initio calculations. While the calculation of these rate coefficients using ab initio data is becoming routine in many cases, the determination of the uncertainties of the rate coefficients are often ignored, sometimes crudely assessed by varying independently just a few of the numerous parameters, and only occasionally studied in detail. In this study, molecular frequencies, barrier heights, well depths, and imaginary frequencies (needed to calculate quantum mechanical tunneling) were automatically perturbed in an uncorrelated fashion. Our Python tool, MEUQ, takes user requests to change all or specified well, barrier, or bimolecular product parameters for a reaction. We propagate the uncertainty in these input parameters and perform global sensitivity analysis in the rate coefficients for the ethyl + O 2 system using state-of-the-art uncertainty quantification (UQ) techniques via Python interface to UQ Toolkit (www.sandia.gov/uqtoolkit). A total of 10,000 sets of rate coefficients were collected after perturbing 240 molecular parameters. With our methodology, sensitive mechanistic steps can be revealed to a modeler in a straightforward manner for identification of significant and negligible influences in bimolecular reactions.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Uncertainty in master-equation-based rate coefficient calculations

Rate coefficients are key quantities in gas phase kinetics and can be determined theoretically via master equation (ME) calculations. Rate coefficients characterize how fast a certain chemical species reacts away due to collisions into a specific product. Some of these collisions are simply transferring energy between the colliding partners, in which case the initial chemical species can undergo a unimolecular reaction: dissociation or isomerization. Other collisions are reactive, and the colliding partners either exchange atoms, these are direct reactions, or form complexes that can themselves react further or get stabilized by deactivating collisions with a bath gas. The input of MEs are molecular parameters: geometries, energies, and frequencies determined from ab initio calculations. While the calculation of these rate coefficients using ab initio data is becoming routine in many cases, the determination of the uncertainties of the rate coefficients are often ignored, sometimes crudely assessed by varying independently just a few of the numerous parameters, and only occasionally studied in detail. In this study, molecular frequencies, barrier heights, well depths, and imaginary frequencies (needed to calculate quantum mechanical tunneling) were automatically perturbed in an uncorrelated fashion. Our Python tool, MEUQ, takes user requests to change all or specified well, barrier, or bimolecular product parameters for a reaction. We propagate the uncertainty in these input parameters and perform global sensitivity analysis in the rate coefficients for the ethyl + O 2 system using state-of-the-art uncertainty quantification (UQ) techniques via Python interface to UQ Toolkit (www . sandia.gov/uqtoolkit). A total of 10,000 sets of rate coefficients were collected after perturbing 240 molecular parameters. With our methodology, sensitive mechanistic steps can be revealed to a modeler in a straightforward manner for identification of significant and negligible influences in bimolecular reactions.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

LA17-V-InverseRadTransportProblem-PD2Kb. Final Project Report

In the inverse radiation transport problem, a radiation signal emanating from an unknown object is used to reconstruct geometric and material parameters of the object , e.g., the mass of special nuclear material (SNM), the isotopic composition of the SNM (e.g., the enrichment of 235U), the thickness of shielding, etc. The inverse radiation transport problem is one of the most fundamental and important in the emergency response mission, and has been addressed in previous NA-22 and NA-42/84 investments by using derivative-free optimization methods. Our project was aimed at (i) consolidating previous work by applying polynomial chaos expansion (PCE) for reducing significantly the number of forward model simulations required by derivative-free optimization methods; (ii) developing a general-purpose framework and stand-alone software for solving the inverse radiation transport problem by applying a new “predictive modeling” methodology (predictive modeling of coupled multi-physics systems, PM_CMPS) which enables the fusing of experimental and computational information; and (iii) developing several benchmarks (for unscattered gamma rays) for demonstrating the quantification of non-Gaussian features (e.g., skewness and kurtosis, which characterize asymmetries and long tails) of response distributions by deterministically and efficiently propagating parameter uncertainties (covariances and higher-order moments, where available) using high-order sensitivities computed by the recently developed comprehensive adjoint sensitivity analysis methodology (C-ASAM). With regard to (i), application of PCE, the most significant accomplishment of this project was an implementation of PCE into INVERSE that significantly reduced computation time for the DiffeRential Evolution Adaptive Metropolis (DREAM) and Differential Evolution (DE) methods. For inverse transport problems with measurements of neutron leakage, neutron multiplication, and neutron-induced gamma-rays, the solution times for the DREAM and DE methods were reduced by factors of 30-300, reducing run times from hours to minutes while maintaining the same accuracy as the standard (non-PCE) DREAM and DE. Speedups were further improved by exploring sparse grid quadrature sets using the Toolkit for Adaptive Stochastic Modeling and Non-Intrusive ApproximatioN (Tasmanian) toolset. An initial exploration of sparse grids indicated that further speedups factors of 5-10 are obtainable. The implementation of PCE in INVERSE is not in a form that users can easily access. With regard to (ii), solving the inverse problem using a predictive modeling paradigm, the most significant accomplishment of this project were the derivation and implementation of adjoint-based second derivatives in neutron and photon transport theory. There are two codes that compute these derivatives. SENSPG computes them for uncollided gamma rays and SENSMG computes them for neutron leakage. SENSMG is open-source code available on github. Other significant accomplishments included a stand-alone code, MULTI-PRED, that implements the predictive modeling paradigm, along with a detailed manual. The code was implemented in the existing INVERSE code for inverse problems, but a head-to-head comparison of inverse methods was not achieved. We also derived and implemented the capability to compute sensitivities for multiplicity measurements into the SENSMG code, and this capability was used in a data assimilation application, but it was not used for solving inverse problems. With regard to (iii), developing benchmarks to demonstrate the importance of high-order sensitivity and uncertainty analysis, the most significant accomplishment of this project were two test problems that showed how important second-order sensitivities are.

07 ISOTOPE AND RADIATION SOURCES↗

CRADA FINAL REPORT FOR CRADA NO. CR-08-001 CEMENTITIOUS BARRIERS PARTNERSHIPS

The Cementitious Barriers Partnership (CBP), led by the Savannah River National Laboratory, has achieved its ultimate purpose to support progress in final treatment and disposal of legacy waste and closure of High-Level Waste (HLW) tanks in the DOE complex. The CBP achieved its goal by developing experimental programs with results that supported the development of a reasonable and credible set of software tools to improve understanding and prediction of the structural, hydraulic and chemical performance of cementitious barriers and waste forms used in nuclear applications. These experimental data and software tools are and will continue to reduce the uncertainties in the current methodologies for assessing cementitious barrier performance and increase the consistency and transparency of the assessment process. The CBP has made a significant impact in reducing risk and uncertainties in the performance assessments that directly impact U.S. Department of Energy (DOE) environmental cleanup and closure programs. The CBP has developed an experimentally-based, integrated modeling tool that serves as a screening tool known as the CBP Software Toolbox, Version 3.0. The CBP products reduce the uncertainties of current methodologies for assessing cementitious barrier performance and increasing the consistency and transparency of the DOE assessment process. This CBP Software Toolbox predicts the hydraulic properties and chemical stability of cement matrix phases and release fluxes of constituents, in response to variable boundary conditions. Therefore, the CBP tools are reducing the uncertainty by coupling multiscale and multi-physics processes, including physical-chemical evolution and transport phenomena applied to heterogeneous materials with changing boundary conditions. In addition, the CBP Experimental and Characterization Program has provided test data to support chemical and physical degradation mechanisms for cementitious barriers. These tools have application to low activity waste forms, high level waste tank closure, D&D and entombment of major nuclear facilities, landfill waste acceptance criteria, and in-situ grouting and immobilization of vadose zone contamination. The CBP also provided: 1) a forum for DOE-NRC technical exchange, 2) material characterization to support performance assessment (PA) predictions, 3) uncertainty reduction in PA predictions, 4) base case performance to improve PA predictions, and 5) improvement in understanding and quantification of moisture and contaminant transport used in PAs. Additional CBP accomplishments include: sponsorship of a national test bed workshop to obtain collaboration in establishing the path forward to obtain actual data supporting future predictions of cementitious barrier performance, and participation in an International Atomic Energy Agency (IAEA) Cooperative Research Project on the use of cementitious barriers for low-level radioactive waste treatment and disposal.

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

A Framework for Deep Learning Emulation of Numerical Models With a Case Study in Satellite Remote Sensing

Numerical models based on physics represent the state of the art in Earth system modeling and comprise our best tools for generating insights and predictions. Despite rapid growth in computational power, the perceived need for higher model resolutions overwhelms the latest generation computers, reducing the ability of modelers to generate simulations for understanding parameter sensitivities and characterizing variability and uncertainty. Thus, surrogate models are often developed to capture the essential attributes of the full-blown numerical models. Recent successes of machine learning methods, especially deep learning (DL), across many disciplines offer the possibility that complex nonlinear connectionist representations may be able to capture the underlying complex structures and nonlinear processes in Earth systems. A difficult test for DL-based emulation, which refers to function approximation of numerical models, is to understand whether they can be comparable to traditional forms of surrogate models in terms of computational efficiency while simultaneously reproducing model results in a credible manner. A DL emulation that passes this test may be expected to perform even better than simple models with respect to capturing complex processes and spatiotemporal dependencies. Here, we examine, with a case study in satellite-based remote sensing, the hypothesis that DL approaches can credibly represent the simulations from a surrogate model with comparable computational efficiency. Our results are encouraging in that the DL emulation reproduces the results with acceptable accuracy and often even faster performance. We discuss the broader implications of our results in light of the pace of improvements in high-performance implementations of DL and the growing desire for higher resolution simulations in the Earth sciences.

Bayesian Deep Learning↗

Stochastic multiscale flux basis for Stokes-Darcy flows

Three algorithms are developed for uncertainty quantification in modeling coupled Stokes and Darcy flows. The porous media may consist of multiple regions with different properties. The permeability is modeled as a non-stationary stochastic variable, with its log represented as a sum of local Karhunen-Loève (KL) expansions. The problem is approximated by stochastic collocation on either tensor-product or sparse grids, coupled with a multiscale mortar mixed finite element method for the spatial discretization. A non-overlapping domain decomposition algorithm reduces the global problem to a coarse scale mortar interface problem, which is solved by an iterative solver, for each stochastic realization. In the traditional domain decomposition implementation, each subdomain solves a local Dirichlet or Neumann problem in every interface iteration. To reduce this cost, two additional algorithms based on deterministic or stochastic multiscale flux basis are introduced. The basis consists of the local flux (or velocity trace) responses from each mortar degree of freedom. It is computed for each subdomain independently before the interface iteration begins. The use of the multiscale flux basis avoids the need for subdomain solves on each iteration. The deterministic basis is computed at each stochastic collocation and used only at this realization. The stochastic basis is formed by further looping over all local realizations of a subdomain's KL region before the stochastic collocation begins. It is used over multiple realizations. Numerical tests are presented to illustrate the performance of the three algorithms, with the stochastic multiscale flux basis showing significant savings in computational cost compared to the other two algorithms.

97 MATHEMATICS AND COMPUTING↗

Correlating electroluminescence characterization and physics-based models of InGaN/GaN LEDs: Pitfalls and open issues

Electroluminescence (EL) characterization of InGaN/GaN light-emitting diodes (LEDs), coupled with numerical device models of different sophistication, is routinely adopted not only to establish correlations between device efficiency and structural features, but also to make inferences about the loss mechanisms responsible for LED efficiency droop at high driving currents. The limits of this investigative approach are discussed here in a case study based on a comprehensive set of current- and temperature-dependent EL data from blue LEDs with low and high densities of threading dislocations (TDs). First, the effects limiting the applicability of simpler (closed-form and/or one-dimensional) classes of models are addressed, like lateral current crowding, vertical carrier distribution nonuniformity, and interband transition broadening. Then, the major sources of uncertainty affecting state-of-the-art numerical device simulation are reviewed and discussed, including (i) the approximations in the transport description through the multi-quantum-well active region, (ii) the alternative valence band parametrizations proposed to calculate the spontaneous emission rate, (iii) the difficulties in defining the Auger coefficients due to inadequacies in the microscopic quantum well description and the possible presence of extra, non-Auger high-current-density recombination mechanisms and/or Auger-induced leakage. In the case of the present LED structures, the application of three-dimensional numerical-simulation-based analysis to the EL data leads to an explanation of efficiency droop in terms of TD-related and Auger-like nonradiative losses, with a C coefficient in the 10{sup −30} cm{sup 6}/s range at room temperature, close to the larger theoretical calculations reported so far. However, a study of the combined effects of structural and model uncertainties suggests that the C values thus determined could be overestimated by about an order of magnitude. This preliminary attempt at uncertainty quantification confirms, beyond the present case, the need for an improved description of carrier transport and microscopic radiative and nonradiative recombination mechanisms in device-level LED numerical models.

36 MATERIALS SCIENCE↗