SEARCH · Engineering Papers

Results for “model exploration”

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.

SourceResearch connectionFrom yearThrough year

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

At least 865 records · Page 48

All sources NASA NTRS DOE OSTI

DOE OSTI2020

Energy Migration Processes in Re(I) MLCT Complexes Featuring a Chromophoric Ancillary Ligand

We present herein the synthesis, structural characterization, electronic structure calculations, and ultrafast and supra-nanosecond photophysical properties of a series of five Re(I) bichromophores exhibiting metal to ligand charge transfer (MLCT) excited states based on the general formula fac -[Re(N ^ N)(CO) 3 (PNI-py)]PF 6, where PNI-py is 4-piperidinyl-1,8-naphthalimidepyridine and N ^ N is a diimine ligand ( Re1 – 5 ), along with their corresponding model chromophores where 4-ethylpyridine was substituted for PNI-py ( Mod1 – 5 ). The diimine ligands used include 1,10-phenanthroline (phen, 1 ), 2,9-dimethyl-4,7-diphenyl-1,10-phenanthroline (bcp, 2 ), 4,4'-di- tert -butyl-2,2'-bipyridine (dtbb, 3 ), 4,4'-diethyl ester-2,2'-bipyridine (deeb, 4 ), and 2,2'-biquinoline (biq, 5 ). In these metal–organic bichromophores, structural modification of the diimine ligand resulted in substantial changes to the observed energy transfer efficiencies between the two chromophores as a result of the variation in 3 MLCT excited-state energies. The photophysical properties and energetic pathways of the model chromophores were explored in parallel to accurately track the changes that arose from introduction of the organic chromophore pendant on the ancillary ligand. All relevant photophysical and energy transfer processes were probed and characterized using time-resolved photoluminescence spectroscopy, ultrafast and nanosecond transient absorption spectroscopy, and time-dependent density functional theory calculations. Of the five bichromophores in this study, four ( Re1 – 4 ) exhibited a thermal equilibrium between the 3 PNI-py and the 3 MLCT excited state, drastically extending the lifetimes of the parent model chromophores.

14 SOLAR ENERGY↗

DOE OSTI2023

Domain-aware Control-oriented Neural Models for Autonomous Underwater Vehicles

Conventional physics-based modeling is a time-consuming bottleneck in control design for complex nonlinear systems like autonomous underwater vehicles (AUVs). In contrast, purely data-driven models, require a large number of observations and lack operational guarantees for safety-critical systems. Data-driven models leveraging available partially characterized dynamics have potential to provide reliable systems models in a typical data-limited scenario for high value complex systems, thereby avoiding months of expensive expert modeling time. In this work we explore this middle-ground between expert-modeled and pure data-driven modeling. We present control-oriented parametric models with varying levels of domain-awareness that exploit known system structure and prior physics knowledge to create constrained deep neural dynamical system models. We employ universal differential equations to construct data-driven blackbox and graybox representations of the AUV dynamics. In addition, we explore a hybrid formulation that explicitly models the residual error related to imperfect graybox models. We compare the prediction performance of the learned models for different distributions of initial conditions and control inputs to assess their suitability for control.

Shaw Cortez, Wenceslao E.↗

DOE OSTI2022

Observability of negative capacitance of a ferroelectric film: Theoretical predictions

Here, we theoretically explore mechanisms that can potentially give rise to the steady-state and transient negative capacitance in a uniaxial ferroelectric film stabilized by a dielectric layer. The analytical expressions for the steady-state capacitance of a single-domain and polydomain states are derived within the Landau-Ginzburg-Devonshire approach and used to study the state stability vs the domain splitting as a function of dielectric layer thickness. Analytical expressions for the critical thickness of the dielectric layer, polarization amplitude, equilibrium domain period, and susceptibility are obtained and corroborated by finite element modeling. Furthermore we explore the possible effects of nonlinear screening by two types of screening charges at the ferroelectric-dielectric interface and show that if at least one of the screening charges is very slow, the total polarization dynamics can exhibit complex time- and voltage-dependent behaviors that can be interpreted as an observable negative capacitance. In this setting, the transient negative capacitance effect is accompanied by almost zero dielectric susceptibility in a wide voltage range and low frequencies. These results may help to elucidate the observation of the transient negative capacitance in thin ferroelectric films, and identify materials systems that can give rise to the behavior.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

NASA NTRS2013

Probabilistic Modeling of the Renal Stone Formation Module

The Integrated Medical Model (IMM) is a probabilistic tool, used in mission planning decision making and medical systems risk assessments. The IMM project maintains a database of over 80 medical conditions that could occur during a spaceflight, documenting an incidence rate and end case scenarios for each. In some cases, where observational data are insufficient to adequately define the inflight medical risk, the IMM utilizes external probabilistic modules to model and estimate the event likelihoods. One such medical event of interest is an unpassed renal stone. Due to a high salt diet and high concentrations of calcium in the blood (due to bone depletion caused by unloading in the microgravity environment) astronauts are at a considerable elevated risk for developing renal calculi (nephrolithiasis) while in space. Lack of observed incidences of nephrolithiasis has led HRP to initiate the development of the Renal Stone Formation Module (RSFM) to create a probabilistic simulator capable of estimating the likelihood of symptomatic renal stone presentation in astronauts on exploration missions. The model consists of two major parts. The first is the probabilistic component, which utilizes probability distributions to assess the range of urine electrolyte parameters and a multivariate regression to transform estimated crystal density and size distributions to the likelihood of the presentation of nephrolithiasis symptoms. The second is a deterministic physical and chemical model of renal stone growth in the kidney developed by Kassemi et al. The probabilistic component of the renal stone model couples the input probability distributions describing the urine chemistry, astronaut physiology, and system parameters with the physical and chemical outputs and inputs to the deterministic stone growth model. These two parts of the model are necessary to capture the uncertainty in the likelihood estimate. The model will be driven by Monte Carlo simulations, continuously randomly sampling the probability distributions of the electrolyte concentrations and system parameters that are inputs into the deterministic model. The total urine chemistry concentrations are used to determine the urine chemistry activity using the Joint Expert Speciation System (JESS), a biochemistry model. Information used from JESS is then fed into the deterministic growth model. Outputs from JESS and the deterministic model are passed back to the probabilistic model where a multivariate regression is used to assess the likelihood of a stone forming and the likelihood of a stone requiring clinical intervention. The parameters used to determine to quantify these risks include: relative supersaturation (RS) of calcium oxalate, citrate/calcium ratio, crystal number density, total urine volume, pH, magnesium excretion, maximum stone width, and ureteral location. Methods and Validation: The RSFM is designed to perform a Monte Carlo simulation to generate probability distributions of clinically significant renal stones, as well as provide an associated uncertainty in the estimate. Initially, early versions will be used to test integration of the components and assess component validation and verification (V&V), with later versions used to address questions regarding design reference mission scenarios. Once integrated with the deterministic component, the credibility assessment of the integrated model will follow NASA STD 7009 requirements.

kidney stones↗

DOE OSTI2025

Anomalous thermal transport in Eshelby twisted van der Waals nanowires

Dislocations in van der Waals (vdW) layered nanomaterials induce strain and structural changes that substantially impact thermal transport. Understanding these effects could enable the manipulation of dislocations for improved thermoelectric and optoelectronic applications, but experimental insights remain limited. Here, in this study, we use synthetic Eshelby twisted vdW GeS nanowires (NWs) with single screw dislocations as a model system to explore the interplay between dislocation-induced structural modifications and lattice thermal conductivity. Our measurements reveal a monoclinic structure stabilized by the dislocation, leading to a substantial drop in thermal conductivity for larger-diameter NWs (70% at room temperature), supported by first-principles calculations. Interestingly, we also find an anomalous enhancement of thermal conductivity with decreasing diameter in twisted NWs, contrary to typical trends in non-twisted GeS NWs. This is attributed to increased conductivity near the NW cores due to compressive strain around the central dislocations, and aligns with a density-functional-theory-informed core–shell model. Our results highlight the critical role of dislocations in thermal conduction, providing fundamental insights for defect and strain engineering in advanced thermal applications.

36 MATERIALS SCIENCE↗

DOE OSTI2021

Metal Removal by a Free Surface Constructed Wetland and Prediction of Metal Bioavailability and Toxicity with Diffusive Gradients in Thin Films (DGT) and Biotic Ligand Model (BLM)

The H-02 constructed wetland is a free water surface wetland to remove copper (Cu) and zinc (Zn) from the industrial wastewater. In this study, we evaluated the performance of the wetland from 2018 to 2019 and coupled the diffusive gradients in thin films (DGTs) and biotic ligand model (BLM) to explore metal speciation and bioavailability in wetland waters. Surface water samples were collected and piston DGTs were deployed in different sites of the wetland. The H-02 wetland functioned well during the sampling period with high removal efficiencies (Cu: 73.8 ± 1.2% and Zn: 75.2 ± 16.0%). In our study, with the assumption that the combination of BLM predicted inorganic metals species, BLM Cu(II) and BLM Zn(II), were the bioavailable and toxic species, DGT-Cu did not correlate to BLM Cu(II) (P = 0.47), but DGT-Zn positively correlated to BLM Zn(II) (R 2 = 0.35, P < 0.001). Compared to the modeling results of BLM, DGT-indicated labile and/or bioavailable Cu included not only free Cu ions and inorganic Cu complexes but also a high percentage of Cu-labile organic matter complexes. DGT-indicated Zn included free Zn ion, inorganic Zn, and only a low percentage of Zn-labile organic matter complexes. Our findings illustrated the appropriate use of passive sampling techniques and geological modeling when biomonitoring could be substituted. Furthermore, the close monitoring of metal concentrations, speciation, and bioavailability helps us understand metal biogeochemistry and metal removal processes and ensure the long-term sustainability of the constructed wetland.

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2022

Exploring New Ways to Classify Industries for Energy Analysis and Modeling

As the US moves closer to embracing a net zero greenhouse gas emissions position, combustion processes outside the power sector are becoming urgent concerns. Industry is an important end user of energy and relies on fossil fuels used directly for process heating and as feedstocks for a diverse range of applications. Fuel and energy use by industry is heterogeneous, meaning that even a single product group can vary broadly in its production routes and associated energy usage. In the US, the North American Industry Classification System (NAICS) serves as the basis for data collection and reporting. In turn, data based on NAICS is the foundation of most US energy modeling. Thus, the effectiveness of NAICS at representing energy use is a limiting condition for plans to improve energy efficiency and alternatives to fossil fuels in industry. Facility-level data to build more detail into heterogeneous sectors is scarce. This work explores alternative classification schemes for industry based on energy use characteristics, and provides a validation of an approach to make facility-level energy use estimates based on publicly available data from the greenhouse gas reporting program. First, several approaches to industrial taxonomies and their usefulness for industrial energy modeling are summarized. Data from Industrial Assessment Centers is analyzed using unsupervised machine learning techniques to detect clusters. Cladistics, an approach from biology, is adapted to energy and process characteristics of industries. A cladogram is presented for evolutionary directions in the iron and steel sector. Cladograms are a promising tool for constructing scenarios and summarizing directions of sectoral innovation. Finally, validation is performed for facility-level energy estimates from the US EPA Greenhouse Gas Reporting Program. This validation assists in making this data source available for use in energy modeling. Together, this work explores alternative approaches for categorizing industries in a way that aids understanding energy use, and presenting pathways for the future.

29 ENERGY PLANNING, POLICY, AND ECONOMY↗

DOE OSTI2021

Biological mechanisms may contribute to soil carbon saturation patterns

Increasing soil organic carbon (SOC) storage is a key strategy to mitigate rising atmospheric CO 2 , yet SOC pools often appear to saturate, or increase at a declining rate, as carbon (C) inputs increase. Soil C saturation is commonly hypothesized to result from the finite amount of reactive mineral surface area available for retaining SOC, and is accordingly represented in SOC models as a physicochemically determined SOC upper limit. However, mineral-associated SOC is largely microbially generated. In this perspective, we present the hypothesis that apparent SOC saturation patterns could emerge as a result of ecological constraints on microbial biomass—for example, via competition or predation—leading to reduced C flow through microbes and a reduced rate of mineral-associated SOC formation as soil C inputs increase. Microbially explicit SOC models offer an opportunity to explore this hypothesis, yet most of these models predict linear microbial biomass increases with C inputs and insensitivity of SOC to input rates. Synthesis of 54 C addition studies revealed constraints on microbial biomass as C inputs increase. Different hypotheses limiting microbial density were embedded in a three-pool SOC model without explicit limits on mineral surface area. As inputs increased, the model demonstrated either no change, linear, or apparently saturating increases in mineral-associated and particulate SOC pools. Taken together, our results suggest that microbial constraints are common and could lead to reduced mineral-associated SOC formation as input rates increase. We conclude that SOC responses to altered C inputs—or any environmental change—are influenced by the ecological factors that limit microbial populations, allowing for a wider range of potential SOC responses to stimuli. Understanding how biotic versus abiotic factors contribute to these patterns will better enable us to predict and manage soil C dynamics.

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2022

MetaText: Compositional Generalization in Deep Language Models

Compositional generalization, the ability to infer and reason using novel combinations of previously encountered entities and structures, is a trait of great utility across a variety of deep learning tasks; it implies a certain ability to reason using consistent and therefore interpretable rules, to understand the constituent parts of its input at multiple levels of granularity, to be robust to spurious differences between semantically equivalent inputs, and more. We seek to understand the degree to which existing natural language models achieve or fall short of compositional generalization across a variety of tasks, whether there are distinct types of compositional generalization in practice, and to identify avenues of intervention through which we can improve models’ compositional generalization ability. After exploring these questions using an array of models, tasks, and potential interventions we see that large, pretrained language models have encountered sufficient training data to account for a variety of fine-grained compositional behavior but still struggle to reason at the level of phrases or larger language structures. Non-intrusive, data-based interventions in the form of augmenting individual sequences with compressed versions of themselves or deriving new examples from induced grammars prove insufficient to encourage greater levels of compositional reasoning, indicating that future work might benefit most from focusing on changes to a model’s inductive bias at the architectural or loss level, or by integrating compositionality-boosting data interventions into the large-scale pretraining process itself.

97 MATHEMATICS AND COMPUTING↗

NASA NTRS1991

Modes of interconnected lattice trusses using continuum models, part 1

This represents a continuing systematic attempt to explore the use of continuum models--in contrast to the Finite Element Models currently universally in use--to develop feedback control laws for stability enhancement of structures, particularly large structures, for deployment in space. We shall show that for the control objective, continuum models do offer unique advantages. It must be admitted of course that developing continuum models for arbitrary structures is no easy task. In this paper we take advantage of the special nature of current Large Space Structures--typified by the NASA-LaRC Evolutionary Model which will be our main concern--which consists of interconnected orthogonal lattice trusses each with identical bays. Using an equivalent one-dimensional Timoshenko beam model, we develop an almost complete continuum model for the evolutionary structure. We do this in stages, beginning only with the main bus as flexible and then going on to make all the appendages also flexible-except for the antenna structure. Based on these models we proceed to develop formulas for mode frequencies and shapes. These are shown to be the roots of the determinant of a matrix of small dimension compared with mode calculations using Finite Element Models, even though the matrix involves transcendental functions. The formulas allow us to study asymptotic properties of the modes and how they evolve as we increase the number of bodies which are treated as flexible. The asymptotics, in fact, become simpler.

Balakrishnan, A. V.↗

DOE OSTI2020

Ionization Yield in Silicon for eV-Scale Electron-Recoil Processes

The development of single charge resolving, macroscopic silicon detectors has opened a window into rare processes at the O(eV) scale. In order to reconstruct the energy of a given event, or model the charge signal obtained for a given amount of energy absorbed by the electrons in a detector, an accurate charge yield model is needed. In this paper we review existing measurements of charge yield in Silicon, focusing in particular on the region below 1 keV. We highlight a calibration gap between 12-50 eV (referred to as the "UV-gap") and employ a phenomenological model of impact ionization to explore the likely charge yield in this energy regime. Finally, we explore the impact of variations in this model on a test case, that of dark matter scattering off electrons, to illustrate the scientific impact of uncertainties in charge yield.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

DOE OSTI2026

Spatial analysis of cell patterning to aid genetic and phenotypic understanding of grass stomatal density: A case study in maize

Biological processes involve complex hierarchies where composite traits result from multiple component traits. However, holistically understanding of how sets of component traits interact to underpin genotype-to-phenotype relationships is generally lacking. Stomatal density (SD) is a tractable model system for exploring how high-throughput phenotyping (HTP) data could be exploited by a new spatial analysis approach to better understand a developmentally and functionally important trait. SD is a composite trait, resulting from various components related to cell identity and size, which are themselves governed by a series of spatio-developmental processes. Data from 192 recombinant inbred lines of maize [Zea mays (L.)] were analyzed by a new stomatal patterning phenotype (SPP) to (1) describe the average spatial probability distribution of the nearest neighboring stomata; (2) derive a core set of component traits related to cell size, cell packing, and positional probabilities; (3) build a structural equation model of component traits underlying SD; and (4) identify stomatal patterning quantitative trait loci (QTL). The core set of SPP-derived traits explained 74% of the variation in SD. Analyzing SPP component traits allowed some loci previously identified as generic SD QTL to be recognized as specific to lateral versus longitudinal elements of stomatal patterning. Therefore, this study highlights how novel insights can be gained by decomposing a composite trait (e.g., SD) into a set of component traits that were present in HTP data but not previously exploited.

59 BASIC BIOLOGICAL SCIENCES↗

DOE OSTI2020

Fundamental problems in astroparticle physics and cosmology (Final Technical Report)

In the standard cosmological model, structure forms when gas cools and condenses to the center of dark matter halos. The onset of structure is governed by rapid star formation and subsequent evolution of the first massive stars. One of the endpoints of stellar evolution is the formation of a black hole. We showed that all black hole merger events must occur after the universe is about 85 million years old. This prediction is important because it means that if a future experiment makes a detection of a black hole merger event at an earlier time then our understanding of cosmology and the onset of structure formation is not fully developed. In addition, we explored dark matter models that have either self-interactions or experience decays. In the first case, self-interactions may lead to the formation of a thick disk that can be co-planar to the Galactic disc due to correlations in angular momentum between dark matter and baryons that are predicted by tidal torque theory. We showed that for heavy dark matter particles, either the self interaction may not be strong enough to solve the small scale structure motivation or a dark disc cannot be present in the Milky Way. In the second-case, we showed that if dark matter experiences decays into two particles, one of which behaves like radiation, then it is possible to relieve the tension that exists between measurements of the local expansion rate. Finally, we developed methods on how to distinguish gamma-ray excesses from promising dark matter targets as compared to conventional astrophysical sources. This is an important toolset that is applicable to any suspected dark matter source.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

DOE OSTI2021

A Secure Learning Control Strategy via Dynamic Camouflaging for Unknown Dynamical Systems under Attacks

This paper presents a secure reinforcement learning (RL) based control method for unknown linear time-invariant cyber-physical systems (CPSs) that are subjected to compositional attacks such as eavesdropping and covert attack. We consider the attack scenario where the attacker learns about the dynamic model during the exploration phase of the learning conducted by the designer to learn a linear quadratic regulator (LQR), and thereafter, use such information to conduct a covert attack on the dynamic system, which we refer to as doubly learning-based control and attack (DLCA) framework. We propose a dynamic camouflaging based attack-resilient reinforcement learning (ARRL) algorithm which can learn the desired optimal controller for the dynamic system, and at the same time, can inject sufficient misinformation in the estimation of system dynamics by the attacker. The algorithm is accompanied by theoretical guarantees and extensive numerical experiments on a consensus multi-agent system and on a benchmark power grid model.

Mukherjee, Sayak↗

DOE OSTI2022

The influence of thermochemistry on the reactivity of propane, the pentane isomers and $n$-heptane in the low temperature regime

In this study, the influence of thermochemistry on the reactivity of fuels at low temperatures (600–1000 K) is studied here. Specifically, the effect of different sets of thermochemistry on chemical model predictions is explored, where various sets are calculated at different levels of theory in addition to recently updated group additivity values. Experimentally measured ignition delay times for propane, the pentane isomers and n-heptane are simulated using NUIGMech1.2 and replacing the thermochemistry of the low-temperature species with the calculated values. For propane, three different thermochemistry sets were calculated, namely CCSD(T)-F12/TZ-F12//B2PLYP-D3/TZ//B2PLYP-D3/TZ (QM1), CCSD(T)-F12/TZ-F12//B2PLYP-D3/TZ//ωB97X-D/TZ (QM2) and B2PLYP-D3/TZ/ωB97X-D/6-31G*//ωB97X-D/6-31G* (QM3). The QM2 results provide parameters to optimize new group additivity (NGA) values which are used to calculate the fourth set of thermochemistry. The model predictions using these four sets are compared to those using NUIGMech1.2 for propane. As the QM1 and QM2 calculations are expensive, the thermochemistry calculated from the QM3 and NGA calculations are used in the pentane isomer and n-heptane models. For all of the models, it is found that the thermochemistry of the species involved in the low-temperature reaction sequence ($RH$, $\dot{R}$, $RO_2H$, $R\dot{O}_2$, $\dot{Q}OOH$ and $\dot{O}_2QOOH$ species) significantly affect fuel reactivity. The NGA values were developed based on all of these species except $\dot{O}_2QOOH$ radicals. The thermochemistry of $\dot{O}_2QOOH$ species cannot be accurately calculated with the NGA representations due to the importance of non-next-nearest neighbor interactions of –OOH substitution. Further development of the NGA method to capture such interactions is in progress. Overall, the model developed using the NGA thermochemistry shows better agreement with experimental data than the model using thermochemistry from affordable and prominent QM methods, such as QM3. Based on the results presented for propane, the pentane isomers and n-heptane, the thermochemistry calculated using the NGA method can be used to model the oxidation of higher order hydrocarbons at low temperatures.

33 ADVANCED PROPULSION SYSTEMS↗

DOE OSTI2025

Simulating Atmospheric Processes in Earth System Models and Quantifying Uncertainties With Deep Learning Multi‐Member and Stochastic Parameterizations

Abstract Deep learning is a powerful tool to represent subgrid processes in climate models, but many application cases have so far used idealized settings and deterministic approaches. Here, we develop stochastic parameterizations with calibrated uncertainty quantification to learn subgrid convective and turbulent processes and surface radiative fluxes of a superparameterization embedded in an Earth System Model (ESM). We explore three methods to construct stochastic parameterizations: (a) a single Deep Neural Network (DNN) with Monte Carlo Dropout; (b) a multi‐member parameterization; and (c) a Variational Encoder Decoder with latent space perturbation. We show that the multi‐member parameterization improves the representation of convective processes, especially in the planetary boundary layer, compared to individual DNNs. The respective uncertainty quantification illustrates that methods (b) and (c) are advantageous compared to a dropout‐based DNN parameterization regarding the spread of convective processes. Hybrid simulations with our best‐performing multi‐member parameterizations remained challenging and crash within the first days. Therefore, we develop a pragmatic partial coupling strategy relying on the superparameterization for condensate emulation. Partial coupling reduces the computational efficiency of hybrid Earth‐like simulations but enables model stability over 5 months with our multi‐member parameterizations. However, our hybrid simulations exhibit biases in thermodynamic fields and differences in precipitation patterns. Despite this, the multi‐member parameterizations enable improvements in reproducing tropical extreme precipitation compared to a traditional convection parameterization. Despite these challenges, our results indicate the potential of a new generation of multi‐member machine learning parameterizations leveraging uncertainty quantification to improve the representation of stochasticity of subgrid effects.

Behrens, Gunnar [Deutsches Zentrum für Luft‐ und R↗

NASA NTRS2013

Validating an Air Traffic Management Concept of Operation Using Statistical Modeling

Validating a concept of operation for a complex, safety-critical system (like the National Airspace System) is challenging because of the high dimensionality of the controllable parameters and the infinite number of states of the system. In this paper, we use statistical modeling techniques to explore the behavior of a conflict detection and resolution algorithm designed for the terminal airspace. These techniques predict the robustness of the system simulation to both nominal and off-nominal behaviors within the overall airspace. They also can be used to evaluate the output of the simulation against recorded airspace data. Additionally, the techniques carry with them a mathematical value of the worth of each prediction-a statistical uncertainty for any robustness estimate. Uncertainty Quantification (UQ) is the process of quantitative characterization and ultimately a reduction of uncertainties in complex systems. UQ is important for understanding the influence of uncertainties on the behavior of a system and therefore is valuable for design, analysis, and verification and validation. In this paper, we apply advanced statistical modeling methodologies and techniques on an advanced air traffic management system, namely the Terminal Tactical Separation Assured Flight Environment (T-TSAFE). We show initial results for a parameter analysis and safety boundary (envelope) detection in the high-dimensional parameter space. For our boundary analysis, we developed a new sequential approach based upon the design of computer experiments, allowing us to incorporate knowledge from domain experts into our modeling and to determine the most likely boundary shapes and its parameters. We carried out the analysis on system parameters and describe an initial approach that will allow us to include time-series inputs, such as the radar track data, into the analysis

Statistical emulation↗

DOE OSTI2024

Machine learning surrogates for surface complexation model of uranium sorption to oxides

Abstract The safety assessments of the geological storage of spent nuclear fuel require understanding the underground radionuclide mobility in case of a leakage from multi-barrier canisters. Uranium, the most common radionuclide in non-reprocessed spent nuclear fuels, is immobile in reduced form (U(IV) and highly mobile in an oxidized state (U(VI)). The latter form is considered one of the most dangerous environmental threats in the safety assessments of spent nuclear fuel repositories. The sorption of uranium to mineral surfaces surrounding the repository limits their mobility. We quantify uranium sorption using surface complexation models (SCMs). Unfortunately, numerical SCM solvers often encounter convergence problems due to the complex nature of convoluted equations and correlations between model parameters. This study explored two machine learning surrogates for the 2-pK Triple Layer Model of uranium retention by oxide surfaces if released as U(IV) in the oxidizing conditions: random forest regressor and deep neural networks. Our surrogate models, particularly DNN, accurately reproduce SCM model predictions at a fraction of the computational cost without any convergence issues. The safety assessment of spent fuel repositories, specifically the migration of leaked radioactive waste, will benefit from having ultrafast AI/ML surrogates for the computationally expensive sorption models that can be easily incorporated into larger-scale contaminant migration models. One such model is presented here.

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗