Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Model validation”

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 289 records · Page 16
All sourcesNASA NTRSDOE OSTI
DOE OSTI2016

Validation of an Analytical Model for Therapeutic and Stray Dose Calculations of 6 MV Photon Beams

Cancer is one of the most significant health care problems today. It is estimated that 1 in 2 men and women will be diagnosed with some form of cancer in their lifetime. Nearly two thirds of cancer patients will receive radiation therapy at some point in their care and the vast majority of these radiotherapy treatments will be performed with megavoltage photon therapy. Photon treatment techniques typically devote considerable attention to limiting the exposure of healthy tissues outside of the intended treatment volume. Nonetheless, small levels of unwanted stray radiation are incident upon the patient's whole body, which can increase the risk for numerous late effects, including second malignancies. When one considers that second cancers account for approximately 15% of all cancers and that it is projected that by the year 2020, there will be almost 20 million cancer survivors in the United States and four million in the UK, the need for improved accuracy in stray dose calculations appears likely to increase dramatically. Despite the obvious benefits this capability would afford, contemporary commercial treatment planning systems (TPSs) lack the ability to accurately predict stray dose outside the treatment field, e.g., beyond the 5% isodose surface. Thus, there is a large, variable, and systematic underestimation of stray radiation exposures by commonly used clinical TPSs for photon therapy. To overcome this problem, in a companion paper we proposed a simple, physics-based analytical model that calculates the total therapeutic and stray dose from photon treatment fields. The purpose of this work was to evaluate the accuracy of the basic physics in the analytical absorbed dose model for several simple treatment fields. We assessed the performance of the basic physics in an analytical model that calculates the total absorbed dose deposited by primary, leakage, and scattered photons both inside and outside of square treatment fields of 6 MV photon conventional radiotherapy. The assessment was accomplished by comparing model predictions of total dose with measured dose profiles in a water-phantom. The major finding of this study is that the physics-based model performed well, predicting dose in water within 11.5% dose difference or 5 mm distance to agreement with measured values in the therapeutic region and within 1.1 mGy/Gy or 0.5 mm distance to agreement with measured values in the out-of-field region at 90% of the locations considered. Contemporary external-beam photon TPSs do not include the capability to accurately predict the absorbed dose beyond a few centimeters outside of the treatment field. Our work suggests that it is feasible to improve dosimetric accuracy in the out-of-field region using a simple and computationally fast, physics-based analytical algorithm. This is demonstrated by the measured absorbed dose as well as the corresponding values calculated by two contemporary commercial TPSs (Pinnacle, Philips Health care, Andover, MA and Eclipse, Varian Medical Systems, Inc., Palo Alto, CA) and the analytical model considered in this work. It is apparent that both TPSs underestimated the absorbed dose by greater than 3 mGy/Gy at locations outside of the 0.5% relative dose level. In contrast, the analytical model evaluated in this work calculated absorbed dose in the same region within 0.4 mGy/Gy of measured values. The physics-based analytical dose model could prove useful in a number of different applications. The accurate knowledge of absorbed dose deposited outside of the treatment field is especially important for predicting radiation late effects, such as second cancer, fertility complications, and cardiac toxicity. It may also facilitate predictions of the effects of in utero exposure. We conclude that it is possible to accurately calculate the therapeutic and stray absorbed dose from a conventional radiotherapy treatment field of 6 MV photons using a physics-based analytical model. This was demonstrated by the agreement between measured and calculated doses in water within 11.5% relative dose difference or 5 mm distance to agreement in the therapeutic region, and within 1.1 mGy/Gy absolute dose difference or 0.5 mm distance to agreement in the out-of-field region at 90% of positions for all of the fields considered in this work. This is also illustrated by Figure 4 which plots dose values from analytical model predictions and measurements at five representative irradiation conditions and shows that the analytical model agreement is good in all cases. Furthermore, the model validated in this work evidently can predict dose values with better accuracy at out-of-field locations than contemporary TPSs. This and previous works, taken together, suggest that analytical physics-based models will offer advantages over TPSs or Monte Carlo models for whole-body dose calculations for several radiotherapy applications.

61 RADIATION PROTECTION AND DOSIMETRY↗
DOE OSTI2019

Validation Tests of Resuspension Models for a Finite and Infinite Site

Dose assessment for deposited radionuclides often requires estimates of air concentrations that are derived from measured soil concentrations. For this, dose assessors typically use literature resuspension values that while empirically based, can vary by orders of magnitude making it difficult to provide accurate dose estimates. Despite the complexities of the physical processes involved in resuspension, the models generally used for dose assessment are relatively simplistic and rarely are the models validated for a specific site, thus making prediction of air concentrations or airborne emissions highly uncertain. Additionally, the size of the contaminated area can have an impact on downwind concentrations, yet literature values do not account for smaller-sized contaminated sites adding additional uncertainty. To test resuspension models for soil-bound radionuclides at finite and infinite spatial scales, measurements of soil and air concentrations are made at 1) a location downwind of a former outfall where plutonium-239 was released into the environment ( a finite site), and 2) uncontaminated locations where regional air sampling provides background measurements of naturally occurring uranium-238 in sampled dust (an infinite site). Measured air concentrations were compared to those predicted using the Resuspension Factor Model and the Mass Loading Model. An Area Factor was applied to the smaller contaminated site to account for dilution of dust from the contaminated site with dust originating from offsite locations. Here, results show that when properly parameterized to site conditions, resuspension models can predict air concentrations to within a factor of ten.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2019

Multiscale Modeling of CO 2 Migration and Trapping in Fractured Reservoirs with Validation by Model Comparison and Real-Site Applications (Final Technical Report)

This report documents the accomplishments achieved during the project titled “Multiscale Modeling of CO 2 Migration and Trapping in Fractured Reservoirs with Validation by Model Comparison and Real-Site Applications” funded by the US Department of Energy, Office of Fossil Energy. The objectives of the project were to develop modeling capabilities for predicting CO 2 and brine migration in fractured reservoirs during geologic carbon storage and to investigate the feasibility of carbon storage in fractured reservoirs using the newly developed modeling capabilities. To achieve these objectives new mass transfer functions were developed to adapt existing dual-continuum approaches – a commonly used approach in hydrocarbon reservoir modeling – to the CO 2 -brine system. In the dual-continuum approach the undisturbed rock matrix and the fractures are modeled as two separate continua, which are coupled through mass exchange of CO 2 and brine between the two continua. Mass transfer functions represent the mass exchange between fractures and the rock matrix, without requiring highly resolved numerical grids of individual fractures, thus allowing large modeling domains relevant to geologic carbon storage questions (tens of meters to hundreds of kilometers). Mass transfer functions for the exchange of both free-phase CO 2 and brine between fractures and rock matrix were developed for both the injection phase and the post-injection phase. During the injection phase the fractures are filled with CO 2 , while the rock matrix is initially filled with brine. Therefore, a mass transfer function was developed that is based on vertical displacement of brine by CO 2 in the rock matrix due to gravitational forces. Once CO 2 injection ceases and CO 2 migrates away from the direct vicinity of the injection site, the fractures will begin to re-fill with brine, while the rock matrix will have high CO 2 saturations. For these conditions a mass transfer functions was developed that is based on spontaneous imbibition, as the pore spaces in the rock matrix are much smaller than in the fractures. For formations with low permeability or high capillary entry pressure of the rock matrix, free-phase CO 2 may not be able to enter the rock matrix. However, free-phase CO 2 at the interface between fractures and rock matrix can dissolve into brine resident in the rock matrix. Therefore, additional mass transfer functions were developed based on aqueous-phase diffusion of CO 2 through the rock matrix. These mass transfer functions are based on existing solutions for diffusion and were expanded to be applicable to a variety of block shapes relevant to dual-continuum modeling. Implementations of the newly developed mass transfer functions showed that they give accurate results when compared to highly resolved models. In addition to developing new transfer functions, a new vertically-integrated dual-continuum modeling framework was developed. This development takes advantage of fast vertical segregation of CO 2 and brine in the fractures due to the density difference between CO 2 and brine and the high permeability of the fractures. Therefore, a vertically-integrated vertical equilibrium model for the fractures is coupled with a vertically-distributed model for the rock matrix. The newly developed approach significantly reduces the computational effort, but is only accurate for fracture permeabilities of 100 mD or higher. Finally, the newly developed models were applied to real and hypothetical geologic carbon storage sites to investigate the feasibility of using fractured reservoirs for carbon storage. The general results were, that storage of CO 2 – both as free-phase and dissolved – is feasible, but that storage capacities are likely lower than in unfractured systems. While the high permeability of the fractures reduces the injection pressure (or permits higher injection rates), fast migration of CO 2 may limit the overall storage volume, because CO 2 may escape laterally with not enough of the rock matrix being swept for effective storage.

20 FOSSIL-FUELED POWER PLANTS↗
DOE OSTI2017

Evaluating the Sources of Uncertainties in the Measurements from Multiple Pyranometers and Pyrheliometers

Traceable radiometric data sets are essential for validating climate models, validating satellite-based models for estimating solar resources, and validating solar radiation forecasts. The current state-of-the-art radiometers have uncertainties in the range from 2% - 5% and sometimes more [1]. The National Renewable Energy Laboratory (NREL) and other organizations are identifying uncertainties and improving radiometric measurement performance and developing a consensus methodology for acquiring radiometric data. This study analyzes the impact of differing specifications -- such as cosine response, thermal offset, spectral response, and others -- on the accuracy of radiometric data for various radiometers. The study will also provide insight on how to perform a measurement uncertainty analysis and how to reduce the impact of some of the sources of uncertainties.

14 SOLAR ENERGY↗
DOE OSTI2024

Wave Energy Converter Power Take-Off Modeling and Validation From Experimental Bench Tests

This article describes the implementation of a new numerical model of the power take-off system installed in the Monterey Bay Aquarium Research Institute wave energy converter, a device developed to provide power to various oceanic research missions. The simultaneous presence of hydraulic, pneumatic, and electrical subsystems in the power take-off system represents a significant challenge in forging an accurate model able to replicate the main dynamic characteristics of the system. The validation of the new numerical model is addressed by comparing simulations with the measurements obtained during a series of bench tests. Data from the bench tests show good agreement with the numerical model. The validated model provides deeper insights into the complex nonlinear dynamics of the power take-off system and will support further performance improvements in the future.

16 TIDAL AND WAVE POWER↗
DOE OSTI2022

Experimental validation of multiphysics model simulations of the thermal response of a cement clinker rotary kiln at laboratory scale

Abstract An increasing demand for buildings, transportation systems and civil infrastructure development has driven expansion of cement consumption world‐wide, producing a significant increase in related global energy demand. With approximately 7% of the world‐wide industrial energy consumption (10.7 exajoules [EJ]), the cement industry is the third most energy intensive industrial processes and a key component for concrete, the most consumed composite material in the global construction industry. In cement manufacturing, the cement kiln accounts for most of the energy consumption in the production process. As the heart of a cement plant, the cement kiln is where the kiln feed primarily containing calcium oxide (CaO), silica (SiO 2 ), alumina (Al 2 O 3 ), and iron (Fe 2 O 3 ) are thermally and chemically transformed into clinker minerals. The presented work developed a multiphysics model, designed and built a laboratory‐scale rotary cement clinker kiln, and produced cement clinker at laboratory‐scale. The model was developed to study the interaction between the various thermal, fluid dynamic and chemical interactions involved in the sintering process used to form Portland cement clinker in an effort to reduce energy use. The analytical model was validated through experimental testing using a unique laboratory‐scale rotary cement kiln developed during the investigation. Also demonstrated was the feasibility of producing clinker at laboratory scale. This modeling and lab scale tests were designed to better understand the clinker sintering process so that operational and quality decisions can be made to optimize energy consumption without compromising cement clinker quality. The computational fluid dynamics modeling was developed in COMSOL Multiphysics 6.0. The characteristics of the combustion fluid flow, concentration of species, temperature and heat transfer were studied for a turbulent flow of methane (CH 4 ) gas and oxygen (O 2 ). Theory suggests that heat transfer impacts the cement production process but the multiphysics model more accurately describes the convection, conduction, and radiant heat transfer in the kilning process and thus allows for a better understanding of the energy exchange driving the chemical reactions that produce Portland cement. Clinker minerals were formed because of appropriate burning conditions implemented during experimental model validation.

Tabares, Juan David↗
DOE OSTI2024

Calibration verification for stochastic agent-based disease spread models

Accurate disease spread modeling is crucial for identifying the severity of outbreaks and planning effective mitigation efforts. To be reliable when applied to new outbreaks, model calibration techniques must be robust. However, current methods frequently forgo calibration verification (a stand-alone process evaluating the calibration procedure) and instead use overall model validation (a process comparing calibrated model results to data) to check calibration processes, which may conceal errors in calibration. In this work, we develop a stochastic agent-based disease spread model to act as a testing environment as we test two calibration methods using simulation-based calibration, which is a synthetic data calibration verification method. The first calibration method is a Bayesian inference approach using an empirically-constructed likelihood and Markov chain Monte Carlo (MCMC) sampling, while the second method is a likelihood-free approach using approximate Bayesian computation (ABC). Simulation-based calibration suggests that there are challenges with the empirical likelihood calculation used in the first calibration method in this context. These issues are alleviated in the ABC approach. Despite these challenges, we note that the first calibration method performs well in a synthetic data model validation test similar to those common in disease spread modeling literature. We conclude that stand-alone calibration verification using synthetic data may benefit epidemiological researchers in identifying model calibration challenges that may be difficult to identify with other commonly used model validation techniques.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2016

Modeling and Validation of Sodium Plugging for Heat Exchangers in Sodium-cooled Fast Reactor Systems

The project “Modeling and Validation of Sodium Plugging for Heat Exchangers in Sodium-cooled Fast Reactor Systems” was conducted jointly by Westinghouse Electric Company (Westinghouse) and Argonne National Laboratory (ANL), over the period October 1, 2013- March 31, 2016. The project’s motivation was the need to provide designers of Sodium Fast Reactors (SFRs) with a validated, state-of-the-art computational tool for the prediction of sodium oxide (Na 2 O) deposition in small-diameter sodium heat exchanger (HX) channels, such as those in the diffusion bonded HXs proposed for SFRs coupled with a supercritical CO 2 (sCO 2 ) Brayton cycle power conversion system. In SFRs, Na 2 O deposition can potentially occur following accidental air ingress in the intermediate heat transport system (IHTS) sodium and simultaneous failure of the IHTS sodium cold trap. In this scenario, oxygen can travel through the IHTS loop and reach the coldest regions, represented by the cold end of the sodium channels of the HXs, where Na 2 O precipitation may initiate and continue. In addition to deteriorating HX heat transfer and pressure drop performance, Na 2 O deposition can lead to channel plugging especially when the size of the sodium channels is small, which is the case for diffusion bonded HXs whose sodium channel hydraulic diameter is generally below 5 mm. Sodium oxide melts at a high temperature well above the sodium melting temperature such that removal of a solid plug such as through dissolution by pure sodium could take a lengthy time. The Sodium Plugging Phenomena Loop (SPPL) was developed at ANL, prior to this project, for investigating Na 2 O deposition phenomena within sodium channels that are prototypical of the diffusion bonded HX channels envisioned for SFR-sCO 2 systems. In this project, a Computational Fluid Dynamic (CFD) model capable of simulating the thermal-hydraulics of the SPPL test section and provided with Na 2 O deposition prediction capabilities, was developed. This state-of-the-art computational tool incorporates a first-principles Na 2 O deposition model developed by ANL, and combines it with predictive capabilities for the spatial and temporal variation of temperature, velocity, dissolved oxygen concentration, and wall temperature under flowing sodium conditions. The CFD model was validated under no-deposition conditions using experimental data collected with the SPPL, demonstrating the model’s capability to predict the thermal-hydraulics of the SPPL test section within the measurement uncertainty characterizing the SPPL instrumentation. The model’s deposition prediction capability was not, however, validated as the SPPL could not be operated under plugging conditions during the project, resulting in the lack of deposition data with adequate pedigree for a CFD model validation. Two novel diagnostic techniques to detect and characterize Na 2 O deposits, i.e. Ultrasonic Time Domain Reflectometry (UTDR) and Potential Drop (PD) techniques, were developed to ultimately assist in the validation effort under plugging conditions, which can be performed once the SPPL becomes operational. This development effort consisted first in demonstrating, analytically and/or computationally, the capability of these techniques to diagnose Na 2 O deposits inside of small channels (particularly the deposit’s thickness), and subsequently in the fabrication and testing of prototypical UTDR and PD instrumentation. The testing, performed on mockups of the SPPL test section, demonstrated the capability of these techniques to detect and characterize material discontinuities like those induced by sodium oxide deposition on stainless steel channel walls. Because of the mentioned impossibility to run the SPPL in a plugging mode, the developed instrumentation could not be tested in-situ, i.e. at the SPPL while deposits are being formed inside of the SPPL test section. Recommended future work includes a possible enhancement in the CFD modeling technique and installation of the developed UTDR and PD instrumentation on the test section, followed by plugging tests to be conducted with the SPPL. The installation of the UTDR and PD diagnostic instrumentation on the SPPL test section will allow collection of Na 2 O deposition data after the onset of deposition to nearly complete channel plugging, which can ultimately be used for the validation of the CFD model.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗
DOE OSTI2022

Thermal modeling of three lakes within the continuous permafrost zone in Alaska using the LAKE 2.0 model

Abstract. Lakes in the Arctic are important reservoirs of heat with much lower albedo in summer and greater absorption of solar radiation than surrounding tundra vegetation. In the winter, lakes that do not freeze to their bed have a mean annual bed temperature >0 ∘C in an otherwise frozen landscape. Under climate warming scenarios, we expect Arctic lakes to accelerate thawing of underlying permafrost due to warming water temperatures in the summer and winter. Previous studies of Arctic lakes have focused on ice cover and thickness, the ice decay process, catchment hydrology, lake water balance, and eddy covariance measurements, but little work has been done in the Arctic to model lake heat balance. We applied the LAKE 2.0 model to simulate water temperatures in three Arctic lakes in northern Alaska over several years and tested the sensitivity of the model to several perturbations of input meteorological variables (precipitation, shortwave radiation, and air temperature) and several model parameters (water vertical resolution, sediment vertical resolution, depth of soil column, and temporal resolution). The LAKE 2.0 model is a one-dimensional model that explicitly solves vertical profiles of water state variables on a grid. We used a combination of meteorological data from local and remote weather stations, as well as data derived from remote sensing, to drive the model. We validated modeled water temperatures with data of observed lake water temperatures at several depths over several years for each lake. Our validation of the LAKE 2.0 model is a necessary step toward modeling changes in Arctic lake ice regimes, lake heat balance, and thermal interactions with permafrost. The sensitivity analysis shows us that lake water temperature is not highly sensitive to small changes in air temperature or precipitation, while changes in shortwave radiation and large changes in precipitation produced larger effects. Snow depth and lake ice strongly affect water temperatures during the frozen season, which dominates the annual thermal regime of Arctic lakes. These findings suggest that reductions in lake ice thickness and duration could lead to more heat storage by lakes and enhanced permafrost degradation.

Clark, Jason A. (ORCID:0000000248105261)↗
DOE OSTI2014

Investigation of Wave Energy Converter Effects on Near-shore Wave Fields: Model Generation Validation and Evaluation - Kaneohe Bay HI.

The numerical model, SWAN (Simulating WAves Nearshore) , was used to simulate wave conditions in Kaneohe Bay, HI in order to determine the effects of wave energy converter ( WEC ) devices on the propagation of waves into shore. A nested SWAN model was validated then used to evaluate a range of initial wave conditions: significant wave heights (H s ) , peak periods (T p ) , and mean wave directions ( MWD) . Differences between wave height s in the presence and absence of WEC device s were assessed at locations in shore of the WEC array. The maximum decrease in wave height due to the WEC s was predicted to be approximately 6% at 5 m and 10 m water depths. Th is occurred for model initiation parameters of H s = 3 m (for 5 m water depth) or 4 m (10 m water depth) , T p = 10 s, and MWD = 330deg . Subsequently, bottom orbital velocities were found to decrease by about 6%.

Roberts, Jesse D.↗
DOE OSTI2022

An experimentally validated mesoscale model for the effective thermal conductivity of U-Zr fuels

An experimentally validated mesoscale model was developed to simulate the effective thermal conductivity (ETC) of metallic U and U–Zr nuclear fuels. The effects of microstructure, temperature, composition, and interfacial thermal resistance (ITR) were investigated. Companion experiments were conducted to validate the model. The numerical simulations clearly demonstrate that accounting for the interface (Kapitza) thermal resistance and Zr precipitation is necessary to improve the model predictions. The dependence of the effective Kapitza resistance of depleted-U and U-10Zr on temperature was determined. Finally, for both materials, the largest difference between the model calculations and the experimental data was about 4.5%, which is within the precision of the experimental measurements.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗
DOE OSTI2024

Validation of Exploding Wire Tests Using Alegra-NG

Exploding wire tests are used to characterize materials undergoing high voltage loading and to validate computational models in a simple configuration. In this report, the materials emphasized are gold, aluminum, copper and nickel wires. Model validation is presented using the nominal model (no uncertainty) in the form of heat maps (per material) that show model bias across scalar quantities and initial voltage. Mesh convergence and uncertainty quantification is also presented.

36 MATERIALS SCIENCE↗
DOE OSTI2016

Robust Ceramic Coaxial Cable Down-Hole Sensors for Long-Term In Situ Monitoring of Geologic CO 2 Injection and Storage

Energy generation by use of fossil fuels produces large volumes of CO 2 and other greenhouse gases, whose accumulation in the atmosphere is widely seen as undesirable. CO 2 Capture followed by sequestration has been identified as the solution. Subsurface geologic formations offer a potential location for long-term storage of CO 2 because of their requisite size. Unfortunately, the inaccessibility and complexity of the subsurface, the wide range of scales of variability, and the coupled nonlinear processes, impose tremendous challenges to determine the transport and predict the fate of the stored CO 2 . Among the various monitoring approaches, in situ down-hole monitoring of the various state parameters provides critical and direct data points that can be used to validate the models, optimize the injection, detect leakage and track the CO 2 plume. However, down-hole sensors that can withstand the harsh conditions and operate over decades of the project lifecycle remain unavailable. Given that the widespread of carbon capture and storage will be the necessity and reality in the future, fundamental and applied research is required to address the significant challenges and technological gaps in lack of long-term reliable down-hole sensors This project focused on the development and demonstration of a novel, low-cost, distributed, robust ceramic coaxial cable sensor platform for in situ down-hole monitoring of geologic CO 2 injection and storage with high spatial and temporal resolutions. The coaxial cable Fabry-Perot interferometer (CCFPI) has been studied as a general sensor platform for in situ, long-term, measurement of temperature, pressure and strain, which are critical to CO 2 injection and storage. A novel signal processing scheme has been developed and demonstrated for dense multiplexing of the sensors for low-cost distributed sensing with high spatial resolution. The developed temperature, pressure and strain sensors have been extensively tested under laboratory conditions that are similar to the downhole CO 2 storage environment, showing excellent capability for in situ monitoring the various parameters that are important to model, optimize the injection, detect leakage and track the CO 2 plume. In addition, the interactions between the sensor datum and the geological models have been investigated in details for the purposes of model validation, guiding sensor installation/placement, enhancement of model prediction capability and optimization of the injection processes. This project has resulted in the successful development of new ceramic coaxial cable based sensor systems that can monitor directly the changes in pressure, temperature, and strain caused by increased reservoir pressure and reduced reservoir temperature due to the supercritical CO 2 injection. Integrated with geological models, the sensors and measurement data can improve the possibility to identify plume movement and leakage in the cap rock and wells with higher precision and more accuracy. The low cost, ease of deployment, small size and dense multiplexing features of the new sensing technology will allow a large number of sensors to be deployed to address the objective to demonstrate that 99% of the CO 2 remains in the injection zone.

01 COAL, LIGNITE, AND PEAT↗
DOE OSTI2026

Uncertainty quantification of fireball features extracted from nuclear test films using computer vision

Films from the US’s historic nuclear testing era comprise the only extensive collection of imagery depicting high-yield detonations. These films offer unique insights into the characteristics of flows occurring on scales that are difficult to replicate experimentally, and they are a valuable source of data for the validation of models used to describe nuclear detonations. In recent work, we implemented modern computer vision and machine learning techniques to extract features of the fireball following nuclear detonation. With a training dataset of fireball films, we fine-tuned a You Only Look Once 11 (YOLO11) model to detect and track the fireball. Applied to a video, the outer bounding box produced in each frame by YOLO11 is used as an input prompt to Meta’s Segment Anything Model 2 (SAM2), which is shown to accurately predict the boundary of the fireball over time with high resolution. These state-of-the-art computer vision foundation models exhibit impressive visual accuracy in their results but lack an output of values that robustly quantify uncertainty in scientific applications. In this paper, we develop procedures for uncertainty quantification of extracted fireball features. We outline the application of a parallel attention mechanism to calculate uncertainty ranges that complement and better pose model validation data. This higher quality fireball validation data may serve to improve prognostic models describing nuclear detonations in support of nuclear forensic and emergency response activities.

Khristy, Joel [ORNL] (ORCID:0000000209963060)↗
DOE OSTI2016

Validation of a cyclic plasticity computational method using fatigue full-field deformation measurements

The evolution of crack tip displacement and strain fields during uniaxial, room temperature, low-cycle fatigue experiments of Nickel superalloy compact tension specimens was measured by a digital image correlation approach and was further used to validate a cyclic plasticity model and corresponding deformation calculations made by a finite elements methodology. The experimental results provided data trends for the opening displacements and near crack tip strains as function of cycles. A finite element model was developed to capture test conditions for a measured crack size. The model captures crack tip plasticity by using a constitutive model calibrated against stress-strain measurements performed on a round bar. Similar quantities were extracted from the model predictions to compare with the digital image correlation measurements for model validation purposes. This type of direct comparison demonstrated that the computational model was capable to adequately capture the crack opening displacements at various stages of the specimen's fatigue life, providing in this way a tool for quantitative cyclic plasticity model validation. In addition, this integrated experimental-computational approach provides a framework to accelerate our understanding related to interactions of fatigue test data and models, as well as ways to inform one another.

36 MATERIALS SCIENCE↗
DOE OSTI2017

Analytical Performance Modeling and Validation of Intel’s Xeon Phi Architecture

Modeling the performance of scientific applications on emerging hardware plays a central role in achieving extreme-scale computing goals. Analytical models that capture the interaction between applications and hardware characteristics are attractive because even a reasonably accurate model can be useful for performance tuning before the hardware is made available. In this paper, we develop a hardware model for Intel’s second-generation Xeon Phi architecture code-named Knights Landing (KNL) for the SKOPE framework. We validate the KNL hardware model by projecting the performance of mini-benchmarks and application kernels. The results show that our KNL model can project the performance with prediction errors of 10% to 20%. The hardware model also provides informative recommendations for code transformations and tuning.

97 MATHEMATICS AND COMPUTING↗