Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Optimization Under Uncertainty”

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 829 records · Page 46
All sourcesNASA NTRSDOE OSTI
DOE OSTI2016

Asynchronous Two-Level Checkpointing Scheme for Large-Scale Adjoints in the Spectral-Element Solver Nek5000

Adjoints are an important computational tool for large-scale sensitivity evaluation, uncertainty quantification, and derivative-based optimization. An essential component of their performance is the storage/recomputation balance in which efficient checkpointing methods play a key role. We introduce a novel asynchronous two-level adjoint checkpointing scheme for multistep numerical time discretizations targeted at large-scale numerical simulations. The checkpointing scheme combines bandwidth-limited disk checkpointing and binomial memory checkpointing. Based on assumptions about the target petascale systems, which we later demonstrate to be realistic on the IBM Blue Gene/Q system Mira, we create a model of the expected performance of our checkpointing approach and validate it using the highly scalable Navier-Stokes spectralelement solver Nek5000 on small to moderate subsystems of the Mira supercomputer. In turn, this allows us to predict optimal algorithmic choices when using all of Mira. We also demonstrate that two-level checkpointing is significantly superior to single-level checkpointing when adjoining a large number of time integration steps. To our knowledge, this is the first time two-level checkpointing had been designed, implemented, tuned, and demonstrated on fluid dynamics codes at large scale of 50k+ cores.

Adjoints↗
DOE OSTI2019

Chapter 6: Modeling and Simulation of Wind-Farm Flows

The subject of this chapter is the modeling and simulation of the flow through full wind farms, also known as wind plants, which are collections of many wind turbines within a region working together to supply power to the grid like any other type of power plant. This topic is interesting and exciting because of the increasing variety of and uses for wind plant flow modeling and simulation. For example, Figure 6.1 shows the computed flow field from a high-fidelity simulation of the Lillgrund offshore wind farm using computational fluid dynamics (CFD) with large-eddy simulation (LES) turbulence modeling and actuator line rotor aerodynamics models. There are many ways to model this type of flow, ranging from very simple models of wind turbine wakes and their interactions that can run in a matter of seconds on a laptop to high-fidelity, turbulence -resolving LES of the atmospheric flow and wakes that require days of use of thousands of cores on a high-performance computing cluster. The uses of wind plant flow simulations are equally varied. Wind plant layout designers run hundreds or thousands of simulations over all possible wind directions and speeds using simple models to predict a future wind plant's annual energy production. Researchers use LESs of the wind plant to study phenomena such as wind turbine wake interactions with each other and with the atmospheric boundary layer (ABL). As computational resources become more readily available, it is even becoming common for turbine manufacturers to use CFD of some form to model wind plants in complex terrain to help in the wind turbine siting and selection process. We are beginning to see the expensive, high-fidelity tools being used to create data and knowledge that improves or enables the creation of new cheaper, lower fidelity tools. We are also beginning to see the low- and high-fidelity tools used together to perform uncertainty quantification of energy estimates and optimization of wind plant layouts. As new computing technologies mature, such as graphics processing unit (GPU)-based computing, the future of full wind plant flow modeling is exciting.

17 WIND ENERGY↗
DOE OSTI2012

COMPREHENSIVE OBSERVATIONS OF THE ULTRAVIOLET SPECTRUM AND IMPROVED ENERGY LEVELS FOR SINGLY IONIZED CHROMIUM (Cr II)

We report new observations of the spectrum of singly ionized chromium (Cr II) in the region 1142-3954 A. The spectra were recorded with the National Institute of Standards and Technology 10.7 m normal-incidence vacuum spectrograph and FT700 vacuum ultraviolet Fourier transform spectrometer. More than 3600 lines are classified as transitions among 283 even and 368 odd levels. The new spectral data are used to re-optimize the energy levels, reducing their uncertainties by a typical factor of 20.

74 ATOMIC AND MOLECULAR PHYSICS↗
DOE OSTI2013

THE SPECTRUM OF Fe II

The spectrum of singly ionized iron (Fe II) has been recorded using high-resolution Fourier transform (FT) and grating spectroscopy over the wavelength range 900 A to 5.5 {mu}m. The spectra were observed in high-current continuous and pulsed hollow cathode discharges using FT spectrometers at the Kitt Peak National Observatory, Tucson, AZ and Imperial College, London and with the 10.7 m Normal Incidence Spectrograph at the National Institute of Standards and Technology. Roughly 12,900 lines were classified using 1027 energy levels of Fe II that were optimized to measured wavenumbers. The wavenumber uncertainties of lines in the FT spectra range from 10{sup -4} cm{sup -1} for strong lines around 4 {mu}m to 0.05 cm{sup -1} for weaker lines around 1500 A. The wavelength uncertainty of lines in the grating spectra is 0.005 A. The ionization energy of (130,655.4 {+-} 0.4) cm{sup -1} was estimated from the 3d{sup 6}({sup 5}D)5g and 3d{sup 6}({sup 5}D)6h levels.

74 ATOMIC AND MOLECULAR PHYSICS↗
DOE OSTI2015

Measurement of the top-quark mass in the tt¯ dilepton channel using the full CDF Run II data set

We present a measurement of the top-quark mass in events containing two leptons (electrons or muons) with a large transverse momentum, two or more energetic jets, and a transverse-momentum imbalance. We use the full proton-antiproton collision data set collected by the CDF experiment during the Fermilab Tevatron Run II at center-of-mass energy √s = 1.96 TeV, corresponding to an integrated luminosity of 9.1 fb –1 . A special observable is exploited for an optimal reduction of the dominant systematic uncertainty, associated with the knowledge of the absolute energy of the hadronic jets. The distribution of this observable in the selected events is compared to simulated distributions of tt¯ dilepton signal and background. We measure a value for the top-quark mass of 171.5±1.9 (stat)±2.5 (syst) GeV/c 2 .

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2019

Joint Chance Constraints Reduction Through Learning in Active Distribution Networks

Due to an increase in distributed generation and controllable loads, distribution networks are frequently operating under high levels of uncertainty. Joint chance constraints, which seek to satisfy multiple constraints simultaneously with a prescribed probability, are one way to incorporate uncertainty across sets of constraints for optimization and control of these networks. Due to the complexity of evaluating these constraints directly, sampling approaches or approximations can be used to transform the joint chance constraint into deterministic constraints. However, sampling techniques may be extremely computationally expensive and not suitable for physical networks operating on fast timescales, and conservative approximations may needlessly result in a much higher cost of system operation. The proposed framework aims to provide a scalable, data-driven approach which learns operational trends in a power network, eliminates zero-probability events (e.g., inactive constraints), and uses this additional information to accurately and efficiently approximate the joint chance constraint directly.

24 POWER TRANSMISSION AND DISTRIBUTION↗
DOE OSTI2016

SU-F-J-192: A Quick and Effective Method to Validate Patient’s Daily Setup and Geometry Changes Prior to Proton Treatment Delivery Based On Water Equivalent Thickness Projection Imaging (WETPI) for Head Neck Cancer (HNC) Patient

Purpose: With the implementation of Cone-beam Computed-Tomography (CBCT) in proton treatment, we introduces a quick and effective tool to verify the patient’s daily setup and geometry changes based on the Water-Equivalent-Thickness Projection-Image(WETPI) from individual beam angle. Methods: A bilateral head neck cancer(HNC) patient previously treated via VMAT was used in this study. The patient received 35 daily CBCT during the whole treatment and there is no significant weight change. The CT numbers of daily CBCTs were corrected by mapping the CT numbers from simulation CT via Deformable Image Registration(DIR). IMPT plan was generated using 4-field IMPT robust optimization (3.5% range and 3mm setup uncertainties) with beam angle 60, 135, 300, 225 degree. WETPI within CTV through all beam directions were calculated. 3%/3mm gamma index(GI) were used to provide a quantitative comparison between initial sim-CT and mapped daily CBCT. To simulate an extreme case where human error is involved, a couch bar was manually inserted in front of beam angle 225 degree of one CBCT. WETPI was compared in this scenario. Results: The average of GI passing rate of this patient from different beam angles throughout the treatment course is 91.5 ± 8.6. In the cases with low passing rate, it was found that the difference between shoulder and neck angle as well as the head rest often causes major deviation. This indicates that the most challenge in treating HNC is the setup around neck area. In the extreme case where a couch bar is accidently inserted in the beam line, GI passing rate drops to 52 from 95. Conclusion: WETPI and quantitative gamma analysis give clinicians, therapists and physicists a quick feedback of the patient’s setup accuracy or geometry changes. The tool could effectively avoid some human errors. Furthermore, this tool could be used potentially as an initial signal to trigger plan adaptation.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2022

MPC4CLR (Model-Predictive-Control-for-Critical-Load-Restoration-in-Power-Distribution-Systems) [SWR-22-24]

Model predictive control (MPC) is a system or process control technique for making decisions under uncertainty via rolling look-ahead optimizations at each control step where only the current step decisions are applied, and the rest are discarded. In this work, we developed an MPC for a critical load restoration (CLR) in power distribution systems to recover system service (electricity delivery) following an extreme event-triggered substation outage. The method considers the problem of controlling distributed energy resources (DERs) of the distribution system with the objective of achieving maximum load pick up while satisfying distribution network flow and voltage constraints. A linearized optimal power flow (OPF) model is employed to represent the physics of the network. The problem formulation is augmented with a ramping (up) reserve product for the DERs to ensure improved and upward monotonic load restoration as time evolves. Simulation analysis and performance tests are performed using a modified IEEE 13-bus test feeder integrated with wind, solar, microturbine, and energy storage battery. The software is developed using various software packages in Julia and Python. The MPC model is implemented using the JuMP optimization language in Julia while the data analytics including renewable generation and load demand forecasts, running the MPC simulation and visualizations is performed in Python.

Eseye, Abinet Tesfaye↗
DOE OSTI2015

Enabling Explosives and Contraband Detection with Neutron Resonant Attenuation. Year 1 of 3 Summary

Material Identification by Resonant Attenuation is a technique that measures the energy-dependent attenuation of 1-10 MeV neutrons as they pass through a sample. Elemental information is determined from the neutron absorption resonances unique to each element. With sufficient energy resolution, these resonances can be used to categorize a wide range of materials, serving as a powerful discrimination technique between explosives, contraband, and other materials. Our proposed system is unique in that it simultaneously down-scatters and time tags neutrons in scintillator detectors oriented between a d-T generator and sample. This allows not only for energy measurements without pulsed neutron beams, but for sample interrogation over a large range of relevant energies, vastly improving scan times. Our system’s core advantage is a potential breakthrough ability to provide detection discrimination of threat materials by their elemental composition (e.g. water vs. hydrogen peroxide) without opening the container. However, several technical and computational challenges associated with this technique have yet to be addressed. There are several open questions: what is the sensitivity to different materials, what scan times are necessary, what are the sources of background, how do each of these scale as the detector system is made larger, and how can the system be integrated into existing scanning technology to close current detection gaps? In order to prove the applicability of this technology, we will develop a validated model to optimize the design and characterize the uncertainties in the measurement, and then test the system in a real-world scenario. This project seeks to perform R&D and laboratory tests that demonstrate proof of concept (TRL 3) to establishing an integrated system and evaluating its performance (TRL 4) through both laboratory tests and a validated detector model. The validated model will allow us to explore our technology’s benefits to explosive detection in various applications.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2016

Enhancements to the RAVEN code in FY16

The RAVEN code has been under development at the Idaho National Laboratory since 2012. Its main goal is to create a multi-purpose platform for the deploying of all the capabilities needed for Probabilistic Risk Assessment, uncertainty quantification, data mining analysis and optimization studies. RAVEN has currently reached a good level of maturity in terms of deployed state-of-art and advanced capabilities. The main subject of this report is to show the activities that have been recently accomplished: • Implementation of ensemble modeling for time-series, and • initial implementation of model validation for surrogate models, and • advanced visualization capability for topology based data analysis The development of ensemble modeling for time-series has been performed in order to begin tackling the needs of those RISMC applications that need to communicate 1-Dimensional information (e.g. power histories, etc.) among different models. In this document the implementation details and an application example is reported. The second subject of this report is about the initial development of methods, within the RAVEN framework, to assess the validity of the predictive capabilities of surrogate models. Indeed, after the construction of a surrogate tight to a certain physical model, it is crucial to assess the goodness of its representation, in order to be confident with its prediction. In this initial activity, a cross-validation technique has been employed. This report will highlight the implementation details and proof its correct implementation by an application example. The final subject of this report is about the implementation of advanced visualization capability in RAVEN, for interactive data analysis. Indeed, RAVEN offers several post-processing capabilities that can structurally decompose data extracted from experimental results offering both data clustering/partitioning and dimensionality reduction techniques. A disadvantage of the workflow available in RAVEN is that it treats these as black box operations and the user is expected to know specific information about their data including the number of partitions to expect in some cases or the “correct” parameter settings for a particular algorithm. In order to overcome this limitation, it has been added an interactive user interface that can be run in RAVEN to explore different parameter settings on the fly specifically for the topological post-processor and allows the user to explore the data interactively. The design of this user interface is generalizable to other algorithms and can be used as a model to generate more dynamic and user-friendly visualization capabilities within RAVEN.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2017

Light Water Reactor Sustainability Program Status of Adaptive Surrogates within the RAVEN framework

The RAVEN code has been under development at the Idaho National Laboratory since 2012. Its main goal is to create a multi-purpose platform for the deploying of all the capabilities needed for Probabilistic Risk Assessment, uncertainty quantification, data mining analysis and optimization studies. RAVEN has demonstrated a good level of maturity in terms of state-of-art and advanced analysis methodologies. Indeed, in the past year the RAVEN code has been released as open-source project and is available to download (raven.inl.gov) free of charge. The main subject of this report is to show the activities that have been recently accomplished with respect the adaptive surrogate modeling: Implementation of the metric system for the assessment of surrogate model validity; Implementation of cross-validation techniques for the global assessment of surrogate models; Implementation of an automatic infrastructure for switching between a high-fidelity model (e.g. physical code) and a surrogate or set of surrogates; Implementation of a scheme for automatic optimization of surrogate model parameters. The aim of this document is to report the status of the development of methods, within the RAVEN framework, to assess the validity of the predictive capabilities of surrogate models. Indeed, after the construction of a surrogate tight to a certain physical model, it is crucial to assess the goodness of its representation, in order to be confident with its prediction. A cross-validation technique has been employed. This report will highlight the implementation details and proof its correct implementation by an application example. The implementation of a way to assess the predictive capabilities of a surrogate allowed the construction of a RAVEN new entity named “HybridModel” that is aimed to switch between an high-fidelity code and its linked surrogate. In addition, leveraging the optimization algorithms that have been implemented in RAVEN last fiscal year, a new scheme has been developed for automatic tuning of the surrogates’ parameters (e.g. penalty factors in support vector machines).

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2019

High-Dimensional Spectral Sampling

A wide variety of applications in science and engineering depend on effectively and efficiently exploring high- dimensional parameter spaces. Common examples are uncertainty quantification, inverse problems, or various optimization approaches. In such applications, the first step is to create an initial uniform, random sampling of the space to create a baseline of knowledge. The simulations evaluated at these samples are then used for subsequent processing. In its most generic form, the goal of sampling is to produce the maximal amount of information with the minimal number of samples. However, despite being the first step and thus influencing the performance of the entire workflow, the quality of the initial sample designs is rarely analyzed or optimized. This project develops a new spectral sampling theory for analyzing and creating experiment designs in high-dimensional spaces. Further, new algorithms for estimating sample properties and sample synthesis are designed. Finally, applications of the proposed spectral sampling tools in surrogate modeling, scientific machine learning and image analysis are presented.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2025

Conceptual Design of a Water Tank Critical Facility for SPARC

Critical experiments, sometimes referred to zero-power reactors, are crucial tools in developing and validating nuclear physics predictions and thus an indispensable capability to ensure criticality safety during all parts of the nuclear fuel cycle and in reducing uncertainties in reactor physics predictions toward optimizing nuclear energy production. The System Physics Advanced Reactor Facility (SPARC) project was recently initiated to enable large-scale criticality experiments using a horizontal split table machine well suited toward solid core materials systems (fuel, moderator, reflector). The facility selected for this mission was a former pool-type research reactor building and thus also well suited toward a second critical experiment capability able to house full-size light-water reactor (LWR) fuel bundles. A conceptual design study was undertaken to review past water tank critical experiments used for LWR physics experiments and to develop an early engineering design for a new critical assembly tank (CAT). The work described here shows that a relatively simple CAT concept can be constructed and deployed in the SPARC facility to meet the urgent demands for new critical experiments on advanced LWR fuel bundles designs. The SPARC facility layout is conducive to the receipt and upending of LWR fuel bundles using existing containers and equipment from the LWR industry. The facility’s overhead crane can then be used to handle fuel bundles and place them in a vertical storage rack or in the CAT for critical experiments, both of which fit within the building’s “open basement” alongside other equipment planned for SPARC and the horizontal split table. A slightly lower area in the basement can serve as a large drain tank so that fail-safe valves drain the CAT reactor tank for safe shutdown. Neutronic configurations were determined where a 3 × 3 array of fuel bundles can be surrounded by full-length “loose rods” to adjust reactivity so that critical is achieved when the bundles are fully submerged. Viable configurations were determined for both pressurized- and boiling-water-reactor-type fuel bundles. This design concept was used to develop an early planning basis for establishing the CAT capability alongside the otherwise planned SPARC project in order to help streamline the process. Recent presidential executive orders have highlighted the need to achieve power uprates in LWR plants and the CAT capability will be a crucial element of these initiatives. Based on the work described herein, it is recommended that an earnest and timely project begin in order to establish this urgently needed capability.

11 - NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗
DOE OSTI2026

User Manual - HydraGNN v5.0: Distributed Implementation of Multi-Tasking Graph Neural Networks

This document serves as the user manual for HydraGNN v5.0, a scalable graph neural network (GNN) architecture for simultaneous prediction of multiple target properties using multi-task learning (MTL). This version of HydraGNN has been developed primarily to support the development, training, and deployment of predictive graph-based deep learning (DL) models for atomistic materials modeling. HydraGNN is templated over 13 message-passing policies, including invariant models (GIN, PNA, PNAPlus, GAT, MFC, CGCNN, SAGE, SchNet, DimeNet) and equivariant models (EGNN, PNAEq, PAINN, MACE), and supports distributed training via distributed data parallelism (DDP), DeepSpeed, and Fully Sharded Data Parallelism (FSDP) on leadership-class supercomputers. Although HydraGNN can be applied to problems beyond atomistic materials modeling, its current use is confined to homogeneous graphs. Additional capabilities include machine-learned interatomic potentials with energy-conserving forces, General, Powerful, and Scalable Graph Transformer (GraphGPS) global attention, periodic boundary conditions, hyperparameter optimization, mixed-precision training, and uncertainty quantification.

97 MATHEMATICS AND COMPUTING↗
NASA NTRS1992

Robust control of flexible space vehicles with minimum structural excitation: On-off pulse control of flexible space vehicles

Both feedback and feedforward control approaches for uncertain dynamical systems (in particular, with uncertainty in structural mode frequency) are investigated. The control objective is to achieve a fast settling time (high performance) and robustness (insensitivity) to plant uncertainty. Preshaping of an ideal, time optimal control input using a tapped-delay filter is shown to provide a fast settling time with robust performance. A robust, non-minimum-phase feedback controller is synthesized with particular emphasis on its proper implementation for a non-zero set-point control problem. It is shown that a properly designed, feedback controller performs well, as compared with a time optimal open loop controller with special preshaping for performance robustness. Also included are two separate papers by the same authors on this subject.

Wie, Bong↗
NASA NTRS1997

Computational Infrastructure for Engine Structural Performance Simulation

Select computer codes developed over the years to simulate specific aspects of engine structures are described. These codes include blade impact integrated multidisciplinary analysis and optimization, progressive structural fracture, quantification of uncertainties for structural reliability and risk, benefits estimation of new technology insertion and hierarchical simulation of engine structures made from metal matrix and ceramic matrix composites. Collectively these codes constitute a unique infrastructure readiness to credibly evaluate new and future engine structural concepts throughout the development cycle from initial concept, to design and fabrication, to service performance and maintenance and repairs, and to retirement for cause and even to possible recycling. Stated differently, they provide 'virtual' concurrent engineering for engine structures total-life-cycle-cost.

Chamis, Christos C.↗