Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “performance modeling”

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 541 records · Page 30

Two-Step Hyperparameter Optimization Method: Accelerating Hyperparameter Search by Using a Fraction of a Training Dataset

Abstract Hyperparameter optimization (HPO) is an important step in machine learning (ML) model development, but common practices are archaic—primarily relying on manual or grid searches. This is partly because adopting advanced HPO algorithms introduces added complexity to the workflow, leading to longer computation times. This poses a notable challenge to ML applications, as suboptimal hyperparameter selections curtail the potential of ML model performance, ultimately obstructing the full exploitation of ML techniques. In this article, we present a two-step HPO method as a strategic solution to curbing computational demands and wait times, gleaned from practical experiences in applied ML parameterization work. The initial phase involves a preliminary evaluation of hyperparameters on a small subset of the training dataset, followed by a reevaluation of the top-performing candidate models postretraining with the entire training dataset. This two-step HPO method is universally applicable across HPO search algorithms, and we argue it has attractive efficiency gains. As a case study, we present our recent application of the two-step HPO method to the development of neural network emulators for aerosol activation. Although our primary use case is a data-rich limit with many millions of samples, we also find that using up to 0.0025% of the data—a few thousand samples—in the initial step is sufficient to find optimal hyperparameter configurations from much more extensive sampling, achieving up to 135× speedup. The benefits of this method materialize through an assessment of hyperparameters and model performance, revealing the minimal model complexity required to achieve the best performance. The assortment of top-performing models harvested from the HPO process allows us to choose a high-performing model with a low inference cost for efficient use in global climate models (GCMs).

97 MATHEMATICS AND COMPUTING↗

Realistic Geochemical Parameter Uncertainty for Performance Assessment Modeling - 17066

Performance assessments (PAs) include uncertain geochemical parameters that affect the fate and transport of contaminants in environmental media, such as partitioning coefficients for water/rock (K{sub d}) and water/air (K{sub H}), diffusion coefficients, and aqueous solubilities. A statistical approach to realistic, as opposed to 'conservative,' stochastic distribution development for geochemical parameters is presented here. For contaminants not identified as focus elements during screening, broad generic distributions are developed. If any parameter is flagged in the subsequent model sensitivity analysis (SA), it becomes a focus element for further research during the next round of distribution development. This iterative approach is efficient and technically defensible at each stage. Data are collected in databases and may be appropriately weighted or filtered. In the relatively few cases where data exist from local or analogous rocks, those data may be prioritized. Once established, the databases can be used for future PAs or other environmental models and can be expanded as needed based on model SA results, additional screening, and as additional relevant data become available. The iterative method is demonstrated for K{sub d} and K{sub H} in a PA model for a radioactive waste disposal facility. An SA was performed based on preliminary geochemical parameter distributions that showed certain model endpoints are sensitive to K{sub d} and K{sub H} values for some contaminants. This underscores the importance of including realistic uncertainty in these geochemical parameters. (authors)

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

Modeling and Predicting Performance of High Performance Computing Applications on Hardware Accelerators

Computers with hardware accelerators, also referred to as hybrid-core systems, speedup applications by offloading certain compute operations that can run faster on accelerators. Thus, it is not surprising that many of top500 supercomputers use accelerators. However, in addition to procurement cost, significant programming and porting effort is required to realize the potential benefit of such accelerators. Hence, before building such a system it is prudent to answer the question 'what is the projected performance benefit from accelerators for the workloads of interest?'. We address this question by way of a performance-modeling framework that predicts realizable application performance on accelerators rapidly and accurately without going to the considerable effort of porting and tuning. The modeling framework first automatically identifies commonly found compute patterns in scientific applications which we term idioms, which may benefit by accelerator technology. Next the framework models the predicted speedup of those idioms if they were to be ported to and run on hardware accelerators. As a proof of concept we characterize two kinds of accelerators 1) the FPGA accelerators on a Convey HC-1 system and 2) an NVIDIA FERMI GPU accelerator. We model performance of the idioms gather/scatter and stream and our predictions show that where these occur in two full-scale HPC applications, Milc and HYCOM, gather/scatter speeds up by as much as 15X, and stream by as much as 14X, whereas the overall compute time of Milc improves by 3.4% and HYCOM by 20%.

Computer Science↗

Efforts to enhance reproducibility in a human performance research project

Background: Ensuring the validity of results from funded programs is a critical concern for agencies that sponsor biological research. In recent years, the open science movement has sought to promote reproducibility by encouraging sharing not only of finished manuscripts but also of data and code supporting their findings. While these innovations have lent support to third-party efforts to replicate calculations underlying key results in the scientific literature, fields of inquiry where privacy considerations or other sensitivities preclude the broad distribution of raw data or analysis may require a more targeted approach to promote the quality of research output. Methods: We describe efforts oriented toward this goal that were implemented in one human performance research program, Measuring Biological Aptitude, organized by the Defense Advanced Research Project Agency's Biological Technologies Office. Our team implemented a four-pronged independent verification and validation (IV&V) strategy including 1) a centralized data storage and exchange platform, 2) quality assurance and quality control (QA/QC) of data collection, 3) test and evaluation of performer models, and 4) an archival software and data repository. Results: Our IV&V plan was carried out with assistance from both the funding agency and participating teams of researchers. QA/QC of data acquisition aided in process improvement and the flagging of experimental errors. Holdout validation set tests provided an independent gauge of model performance. Conclusions: In circumstances that do not support a fully open approach to scientific criticism, standing up independent teams to cross-check and validate the results generated by primary investigators can be an important tool to promote reproducibility of results.

59 BASIC BIOLOGICAL SCIENCES↗

Multi-Year Lags between Forest Browning and Soil Respiration at High Northern Latitudes

High-latitude northern ecosystems are experiencing rapid climate changes, and represent a large potential climate feedback because of their high soil carbon densities and shifting disturbance regimes. A significant carbon flow from these ecosystems is soil respiration (RS, the flow of carbon dioxide, generated by plant roots and soil fauna, from the soil surface to atmosphere), and any change in the high-latitude carbon cycle might thus be reflected in RS observed in the field. This study used two variants of a machine-learning algorithm and least squares regression to examine how remotely-sensed canopy greenness (NDVI), climate, and other variables are coupled to annual RS based on 105 observations from 64 circumpolar sites in a global database. The addition of NDVI roughly doubled model performance, with the best-performing models explaining ~62% of observed RS variability

carbon cycling↗

Application of Machine Learning and Data Augmentation Algorithms in the Discovery of Metal Hydrides for Hydrogen Storage

The development of efficient and sustainable hydrogen storage materials is a key challenge for realizing hydrogen as a clean and flexible energy carrier. Among various options, metal hydrides offer high volumetric storage density and operational safety, yet their application is limited by thermodynamic, kinetic, and compositional constraints. In this work, we investigate the potential of machine learning (ML) to predict key thermodynamic properties—equilibrium plateau pressure, enthalpy, and entropy of hydride formation—based solely on alloy composition using Magpie-generated descriptors. We significantly expand an existing experimental dataset from ~400 to 806 entries and assess the impact of dataset size and data augmentation, using the PADRE algorithm, on model performance. Models including Support Vector Machines and Gradient Boosted Random Forests were trained and optimized via grid search and cross-validation. Results show a marked improvement in predictive accuracy with increased dataset size, while data augmentation benefits are limited to smaller datasets and do not improve accuracy in underrepresented pressure regimes. Furthermore, clustering and cross-validation analyses highlight the limited generalizability of models across different material classes, though high accuracy is achieved when training and testing within a single hydride family (e.g., AB2). The study demonstrates the viability and limitations of ML for accelerating hydride discovery, emphasizing the importance of dataset diversity and representation for robust property prediction.

augmentation↗

Component model reduction via the projection and assembly method

The problem of acquiring a simple but sufficiently accurate model of a dynamic system is made more difficult when the dynamic system of interest is a multibody system comprised of several components. A low order system model may be created by reducing the order of the component models and making use of various available multibody dynamics programs to assemble them into a system model. The difficulty is in choosing the reduced order component models to meet system level requirements. The projection and assembly method, proposed originally by Eke, solves this difficulty by forming the full order system model, performing model reduction at the the system level using system level requirements, and then projecting the desired modes onto the components for component level model reduction. The projection and assembly method is analyzed to show the conditions under which the desired modes are captured exactly; to the numerical precision of the algorithm.

Bernard, Douglas E.↗

Automated clustering-based workload characterization

The demands placed on the mass storage systems at various federal agencies and national laboratories are continuously increasing in intensity. This forces system managers to constantly monitor the system, evaluate the demand placed on it, and tune it appropriately using either heuristics based on experience or analytic models. Performance models require an accurate workload characterization. This can be a laborious and time consuming process. It became evident from our experience that a tool is necessary to automate the workload characterization process. This paper presents the design and discusses the implementation of a tool for workload characterization of mass storage systems. The main features of the tool discussed here are: (1)Automatic support for peak-period determination. Histograms of system activity are generated and presented to the user for peak-period determination; (2) Automatic clustering analysis. The data collected from the mass storage system logs is clustered using clustering algorithms and tightness measures to limit the number of generated clusters; (3) Reporting of varied file statistics. The tool computes several statistics on file sizes such as average, standard deviation, minimum, maximum, frequency, as well as average transfer time. These statistics are given on a per cluster basis; (4) Portability. The tool can easily be used to characterize the workload in mass storage systems of different vendors. The user needs to specify through a simple log description language how the a specific log should be interpreted. The rest of this paper is organized as follows. Section two presents basic concepts in workload characterization as they apply to mass storage systems. Section three describes clustering algorithms and tightness measures. The following section presents the architecture of the tool. Section five presents some results of workload characterization using the tool.Finally, section six presents some concluding remarks.

Pentakalos, Odysseas I.↗

Development of a GPU-based high-performance radiative transfer model for the Infrared Atmospheric Sounding Interferometer (IASI)

Satellite-observed radiance is a nonlinear functional of surface properties and atmospheric temperature and absorbing gas profiles as described by the radiative transfer equation (RTE). In the era of hyperspectral sounders with thousands of high-resolution channels, the computation of the radiative transfer model becomes more time-consuming. The radiative transfer model performance in operational numerical weather prediction systems still limits the number of channels we can use in hyperspectral sounders to only a few hundreds. To take the full advantage of such high-resolution infrared observations, a computationally efficient radiative transfer model is needed to facilitate satellite data assimilation. In recent years the programmable commodity graphics processing unit (GPU) has evolved into a highly parallel, multi-threaded, many-core processor with tremendous computational speed and very high memory bandwidth. The radiative transfer model is very suitable for the GPU implementation to take advantage of the hardware's efficiency and parallelism where radiances of many channels can be calculated in parallel in GPUs. In this paper, we develop a GPU-based high-performance radiative transfer model for the Infrared Atmospheric Sounding Interferometer (IASI) launched in 2006 onboard the first European meteorological polar-orbiting satellites, METOP-A. Each IASI spectrum has 8461 spectral channels. The IASI radiative transfer model consists of three modules. The first module for computing the regression predictors takes less than 0.004% of CPU time, while the second module for transmittance computation and the third module for radiance computation take approximately 92.5% and 7.5%, respectively. Our GPU-based IASI radiative transfer model is developed to run on a low-cost personal supercomputer with four GPUs with total 960 compute cores, delivering near 4 TFlops theoretical peak performance. By massively parallelizing the second and third modules, we reached 364x speedup for 1 GPU and 1455x speedup for all 4 GPUs, both with respect to the original CPU-based single-threaded Fortran code with the -O{sub 2} compiling optimization. The significant 1455x speedup using a computer with four GPUs means that the proposed GPU-based high-performance forward model is able to compute one day's amount of 1,296,000 IASI spectra within nearly 10 min, whereas the original single CPU-based version will impractically take more than 10 days. This model runs over 80% of the theoretical memory bandwidth with asynchronous data transfer. A novel CPU-GPU pipeline implementation of the IASI radiative transfer model is proposed. The GPU-based high-performance IASI radiative transfer model is suitable for the assimilation of the IASI radiance observations into the operational numerical weather forecast model.

97 MATHEMATICS AND COMPUTING↗

Aboveground biomass density models for NASA’s Global Ecosystem Dynamics Investigation (GEDI) lidar mission

NASA's Global Ecosystem Dynamics Investigation (GEDI) is collecting spaceborne full waveform lidar data with a primary science goal of producing accurate estimates of forest aboveground biomass density (AGBD). This paper presents the development of the models used to create GEDI's footprint-level (~25 m) AGBD (GEDI04_A) product, including a description of the datasets used and the procedure for final model selection. The data used to fit our models are from a compilation of globally distributed spatially and temporally coincident field and airborne lidar datasets, whereby we simulated GEDI-like waveforms from airborne lidar to build a calibration database. We used this database to expand the geographic extent of past waveform lidar studies, and divided the globe into four broad strata by Plant Functional Type (PFT) and six geographic regions. GEDI's waveform-to-biomass models take the form of parametric Ordinary Least Squares (OLS) models with simulated Relative Height (RH) metrics as predictor variables. From an exhaustive set of candidate models, we selected the best input predictor variables, and data transformations for each geographic stratum in the GEDI domain to produce a set of comprehensive predictive footprint-level models. We found that model selection frequently favored combinations of RH metrics at the 98th, 90th, 50th, and 10th height above ground-level percentiles (RH98, RH90, RH50, and RH10, respectively), but that inclusion of lower RH metrics (e.g. RH10) did not markedly improve model performance. Second, forced inclusion of RH98 in all models was important and did not degrade model performance, and the best performing models were parsimonious, typically having only 1-3 predictors. Third, stratification by geographic domain (PFT, geographic region) improved model performance in comparison to global models without stratification. Fourth, for the vast majority of strata, the best performing models were fit using square root transformation of field AGBD and/or height metrics. There was considerable variability in model performance across geographic strata, and areas with sparse training data and/or high AGBD values had the poorest performance. These models are used to produce global predictions of AGBD, but will be improved in the future as more and better training data become available.

54 ENVIRONMENTAL SCIENCES↗

Aboveground biomass density models for NASA’s Global Ecosystem Dynamics Investigation (GEDI) lidar mission

NASA’s Global Ecosystem Dynamics Investigation (GEDI) is collecting spaceborne full waveform lidar data with a primary science goal of producing accurate estimates of forest aboveground biomass density (AGBD). This paper presents the development of the models used to create GEDI’s footprint-level (~25 m) AGBD (GEDI04_A) product, including a description of the datasets used and the procedure for final model selection. The data used to fit our models are from a compilation of globally distributed spatially and temporally coincident field and airborne lidar datasets, whereby we simulated GEDI-like waveforms from airborne lidar to build a calibration database. We used this database to expand the geographic extent of past waveform lidar studies, and divided the globe into four broad strata by Plant Functional Type (PFT) and six geographic regions. GEDI’s waveform-to-biomass models take the form of parametric Ordinary Least Squares (OLS) models with simulated Relative Height (RH) metrics as predictor variables. From an exhaustive set of candidate models, we selected the best input predictor variables, and data transformations for each geographic stratum in the GEDI domain to produce a set of comprehensive predictive footprint-level models. We found that model selection frequently favored combinations of RH metrics at the 98th, 90th, 50th, and 10th height above ground-level percentiles (RH98, RH90, RH50, and RH10, respectively), but that inclusion of lower RH metrics (e.g. RH10) did not markedly improve model performance. Second, forced inclusion of RH98 in all models was important and did not degrade model performance, and the best performing models were parsimonious, typically having only 1-3 predictors. Third, stratification by geographic domain (PFT, geographic region) improved model performance in comparison to global models without stratification. Fourth, for the vast majority of strata, the best performing models were fit using square root transformation of field AGBD and/or height metrics. There was considerable variability in model performance across geographic strata, and areas with sparse training data and/or high AGBD values had the poorest performance. These models are used to produce global predictions of AGBD, but will be improved in the future as more and better training data become available.

Laura Duncanson↗

Understanding the Impact of Unobservable Variables on the Performance of Predictive Models: The Need for Feature Space Partitioning and Fusion

When developing predictive models over a dataset, the model is globally optimized across the entire feature space to learn a decision boundary. However, when unobservable variables—which cannot be measured or estimated—interact with the observable variables, this can negatively impact the optimization applied to the decision boundary since the data samples introduced by unobservable variables may have little to no association with the applied global optimization. This, consequently, penalizes the entire decision boundary and model performance. This paper examines some of the detrimental effects of unobservable variables, particularly their role in creating new modes in the distribution of observable variables and reducing the separability of class distributions. Such challenges result in skewed or warped decision boundaries and decreased accuracy of model predictions, particularly for interpretable models like logistic regression and decision trees. Through two illustrative case examples, we highlight the need to address the challenges imposed by unobservable variables. We propose a strategy to mitigate these challenges by creating local regions within the feature space through partitioning. This enables the optimization of local models within the regions to overcome the impact of unobservability in different feature space localities. Research into a more sophisticated partitioning strategy and where the partition should be relative to the sample of interest is left as future work. Through the analysis of the impact of unobservability and the development of a partitioning method, we demonstrate the clear need for a partitioning strategy that integrates knowledge from multiple local models to estimate risk factors using information fusion. Thus, we establish the foundation and motivation for using partitioning and information fusion to overcome the effects of unobservability in predictive models. Formal fusion methods, such as Dempster-Shafer theory, can better leverage the information from local regions to improve the performance of interpretable predictive models in the presence of unobservable variables.

Time Series Data↗

Modeling the Performance of Large-Scale CO2 Storage Systems: A Comparison of Different Sensitivity Analysis Methods

NRAP is an initiative within DOE’s Office of Fossil Energy and is led by the National Energy Technology Laboratory (NETL). It is a multi-national-lab effort that leverages broad technical capabilities across the DOE complex to develop an integrated science base that can be applied to risk assessment for long-term storage of carbon dioxide (CO2). NRAP involves five DOE national laboratories: NETL-RUA, Lawrence Berkeley National Laboratory (LBNL), Lawrence Livermore National Laboratory (LLNL), Los Alamos National Laboratory (LANL), and Pacific Northwest National Laboratory (PNNL). The NETL-RUA is an applied research collaboration that combines NETL’s energy research expertise in the Office of Research and Development (ORD) with the broad capabilities of five nationally recognized, regional universities.

42 ENGINEERING↗

Computational Models of Human Performance: Validation of Memory and Procedural Representation in Advanced Air/Ground Simulation

The Man-Machine Interaction Design and Analysis System (MIDAS) under joint U.S. Army and NASA cooperative is intended to assist designers of complex human/automation systems in successfully incorporating human performance capabilities and limitations into decision and action support systems. MIDAS is a computational representation of multiple human operators, selected perceptual, cognitive, and physical functions of those operators, and the physical/functional representation of the equipment with which they operate. MIDAS has been used as an integrated predictive framework for the investigation of human/machine systems, particularly in situations with high demands on the operators. We have extended the human performance models to include representation of both human operators and intelligent aiding systems in flight management, and air traffic service. The focus of this development is to predict human performance in response to aiding system developed to identify aircraft conflict and to assist in the shared authority for resolution. The demands of this application requires representation of many intelligent agents sharing world-models, coordinating action/intention, and cooperative scheduling of goals and action in an somewhat unpredictable world of operations. In recent applications to airborne systems development, MIDAS has demonstrated an ability to predict flight crew decision-making and procedural behavior when interacting with automated flight management systems and Air Traffic Control. In this paper, we describe two enhancements to MIDAS. The first involves the addition of working memory in the form of an articulatory buffer for verbal communication protocols and a visuo-spatial buffer for communications via digital datalink. The second enhancement is a representation of multiple operators working as a team. This enhanced model was used to predict the performance of human flight crews and their level of compliance with commercial aviation communication procedures. We show how the data produced by MIDAS compares with flight crew performance data from full mission simulations. Finally, we discuss the use of these features to study communication issues connected with aircraft-based separation assurance.

Corker, Kevin M.↗

Models used to assess the performance of photovoltaic systems.

This report documents the various photovoltaic (PV) performance models and software developed and utilized by researchers at Sandia National Laboratories (SNL) in support of the Photovoltaics and Grid Integration Department. In addition to PV performance models, hybrid system and battery storage models are discussed. A hybrid system using other distributed sources and energy storage can help reduce the variability inherent in PV generation, and due to the complexity of combining multiple generation sources and system loads, these models are invaluable for system design and optimization. Energy storage plays an important role in reducing PV intermittency and battery storage models are used to understand the best configurations and technologies to store PV generated electricity. Other researcher's models used by SNL are discussed including some widely known models that incorporate algorithms developed at SNL. There are other models included in the discussion that are not used by or were not adopted from SNL research but may provide some benefit to researchers working on PV array performance, hybrid system models and energy storage. The paper is organized into three sections to describe the different software models as applied to photovoltaic performance, hybrid systems, and battery storage. For each model, there is a description which includes where to find the model, whether it is currently maintained and any references that may be available. Modeling improvements underway at SNL include quantifying the uncertainty of individual system components, the overall uncertainty in modeled vs. measured results and modeling large PV systems. SNL is also conducting research into the overall reliability of PV systems.

14 SOLAR ENERGY↗

Practical factors of envelope model setup and their effects on the performance of model predictive control for building heating, ventilating, and air conditioning systems

Model predictive control (MPC) for buildings is attracting significant attention in research and industry due to its potential to address a number of challenges facing the building industry, including energy cost reduction, grid integration, and occupant connectivity. However, the strategy has not yet been implemented at any scale, largely due to the significant effort required to configure and calibrate the model used in the MPC controller. While many studies have focused on methods to expedite model configuration and improve model accuracy, few have studied the impact a wide range of factors have on the accuracy of the resulting model. In addition, few have continued on to analyze these factors’ impact on MPC controller performance in terms of final operating costs. Therefore, this study first identifies the practical factors affecting model setup, specifically focusing on the thermal envelope. The seven that are identified are building design, model structure, model order, data set, data quality, identification algorithm and initial guesses, and software tool-chain. Then, through a large number of trials, it analyzes each factor's influence on model accuracy, focusing on grey-box models for a single zone building envelope. Finally, this study implements a subset of the models identified with these factor variations in heating, ventilating, and air conditioning MPC controllers, and tests them in simulation of a representative case that aims to optimally cool a single-zone building with time-varying electricity prices. It is found that a difference of up to 20%25 in cooling cost for the cases studied can occur between the best performing model and the worst performing model. The primary factors attributing to this were model structure and initial parameter guesses during parameter estimation of the model.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Performance Comparison of Machine Learning Models for Ultrasonic Nondestructive Evaluation of Alkali-Silica Reaction in Concrete

Alkali-silica reaction (ASR) causes concrete degradation, leading to cracking, rebar corrosion, and reduced structural integrity, which raises safety concerns. Ultrasonic nondestructive evaluation (NDE) effectively assesses concrete properties and monitors ASR progression. However, its deployment and analysis require specialized expertise and subjective interpretation. As computational power increases, artificial intelligence (AI) and machine learning (ML) algorithms are increasingly being used to automate NDE data analysis across various industries for AI-assisted automation. Regulatory agencies are adapting to this technological shift, prompting a need to evaluate current ML technologies’ capabilities and limitations in assessing concrete material properties and damage. This report presents a comparative analysis of four ML regression models for predicting concrete material damage induced by ASR expansion using long-term ultrasonic data monitoring. The models investigated include linear regression (LR), support vector regression (SVR), shallow neural networks (NN), and deep neural networks (DNN). LR, SVR, and shallow NN models use features extracted from ultrasonic signals, whereas the DNN model processes time-domain ultrasonic signals and frequency spectra directly. The study systematically compared the models’ performance from various perspectives, including model input, prediction performance, and generalization ability. The findings indicate significant variability in model performance, with some ML algorithms achieving very high or very low prediction accuracy depending on the preprocessing and feature engineering (extraction and selection) applied. Key insights include the observation that shallow ML models (LR, SVR, and shallow NNs) require meticulous preprocessing and feature extraction to achieve high accuracy. In contrast, the DNN model, although it bypasses the need for feature engineering, necessitates extensive preprocessing to mitigate noise and computational demands. The SVR model emerged as the top performer among the shallow models, and the DNN model exhibited superior performance on specific datasets but struggled with generalization across specimens from different batches. Additionally, the SVR model is sensitive to temperature variations, whereas the DNN model is robust in this regard. Using recurrent neural networks is recommended for future ASR expansion prediction studies. Recurrent neural networks’ inherent ability to capture temporal dependencies and long-term patterns makes them well suited for analyzing sequential ultrasonic monitoring data. Overall, the results and conclusions of this study could provide insights into the capabilities and effectiveness of ML when applied to ultrasonic NDE data and help identify best practices for using ML for ultrasonic NDE of concrete material properties.

36 MATERIALS SCIENCE↗

Development of Waste Inventory for Probabilistic Performance Assessment Modeling - 19361

Estimation of radioactive waste inventory is a key initial step in the development of an environmental fate and transport model for performance assessment (PA) of radioactive waste sites. PA model output is often sensitive to parameters that represent inventory as a source term, yet estimation is a difficult problem. Inventory estimation for a probabilistic performance assessment (PPA) is related to, but distinct from, that performed for waste excavation, characterization, and offsite disposal. The latter appropriately focuses on obtaining the most detailed information possible on specific hazards and locations; however, these questions cannot be fully answered without intrusive site characterization efforts. This type of characterization is often not feasible when assembling information regarding a source term in a PPA model. Development of waste inventory for a PPA is one piece of a larger hierarchical model. The physical process model represented by Goldsim computer code represents one level of the hierarchy, and the second level of the hierarchy involves estimation of inputs such as inventory, transport, and risk parameters. The estimation of inputs is not a straightforward problem. The parameters of a computer model often do not have an equivalence to the physical quantities they describe, and estimation of the physical quantities themselves can be difficult due to limitations of available information. Inventory information for the West Valley Site in upstate New York is based on data from historical shipping and disposal forms. In the 1990's, the West Valley Demonstration Project (WVDP) compiled information from existing databases and historical waste records into one complete modern database. A conceptual model was then applied and used with recorded information to characterize the waste mass and radionuclide composition for each waste burial. Information was ultimately aggregated over all waste records to estimate the mass of each radionuclide within each trench, hole, or caisson. Three decades later, we revisit the master database, make updates with newly-identified information, and incorporate the conceptual model previously used for inventory estimation in the hierarchical modeling framework. We characterize uncertainty using probability distributions for waste mass and radionuclide composition at the individual shipment or individual burial level. Uncertainty incorporated at the lowest level of the hierarchy is ultimately carried through to uncertainty in PPA model endpoints. Sensitivity analysis of early model results is then applied to identify influential inputs and guide subsequent efforts towards parameters that drive risk. (authors)

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗