Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “optimal accuracy”

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 1,009 records · Page 56

Explanation of Anomalous Behavior Observed in Impedance Eduction Techniques Using Measured Data

Several enhancements that improve the accuracy and robustness of an impedance eduction technique that use an automatic optimizer are presented. These enhancements are then used to launch an intensive investigation into the cause of anomalous behavior that occurs for a small number of test conditions. This anomalous behavior is investigated for both a hardwall insert and a conventional liner. The primary conclusions of the study are that: (1) for the hard wall insert, the anomalies are due to narrow peaks in the objective function, (2) For the conventional liner, the anomalies are due to the presence of an extremely flat objective function, and (3) the anomalies appear to be triggered by inconsistencies between the duct propagation model and the measured data. At high frequencies, the duct propagation model may need to include the effects of higher-order duct modes, whereas at low frequencies, the effects of the mean boundary layer may have to be included.

Watson, Willie R.↗

Wall-modeled LES of the Three-dimensional Speed Bump Experiment

We evaluate the performance of wall-modeled large eddy simulation (WMLES) in predicting smooth-body turbulent flow separation for a three-dimensional Gaussian-shaped speed bump geometry. The Reynolds number based on the bump length is 2 million, and an unstructured compressible finite-volume solver is used with an equilibrium wall model and dynamic subgrid scale model. Spanwise periodic simulations of the centerline two-dimensional bump were used to assess grid resolution requirements, and resolving the thin internal layer in the accelerating region was found to be necessary to correctly capture the downstream separated flow region. Based on these insights, an optimized grid that was smaller by a factor of two provided comparable accuracy to a finer grid that has been used by us and other researchers in past studies. Using insights gained from the spanwise periodic simulation, an unstructured polyhderal grid with about 250 million cells was used for the three-dimensional (3D) configuration with inviscid tunnel side and top walls. Detailed comparisons of wall skin-friction coefficient, wall pressure, velocity and turbulent stresses with available experimental data indicated excellent agreement. While the low-Reynolds Number Spalart-Allmaras RANS model with rotation/curvature correction gave qualitatively good agreement with experiments, WMLES showed significantly more accurate quantitative predictions of the flowfield. The 3D WMLES showed good agreement with experiments in the separated region, and the centerplane results indicated a different separation topology compared to the spanwise periodic simulation.

Computational Fluid Dynamics↗

Evaluating the feasibility of using downwind methods to quantify point source oil and gas emissions using continuously monitoring fence-line sensors

The dependable reporting of methane (CH 4 ) emissions from point sources, such as fugitive leaks from oil and gas infrastructure, is important for profit maximization (retaining more hydrocarbons), evaluating climate impacts, assessing CH 4 fees for regulatory programs, and validating CH 4 intensity in differentiated gas programs. Currently, there are disagreements between emissions reported by different quantification techniques for the same sources. It has been suggested that downwind CH 4 quantification methods using CH 4 measurements on the fence line of production facilities could be used to generate emission estimates from oil and gas operations at the site level, but it is currently unclear how accurate the quantified emissions are. To investigate the accuracy of downwind methods, this study uses fence-line simulated data collected during controlled-release experiments as input for a non-standard closed-path eddy covariance (EC), the Gaussian plume inverse model (GPIM), and the backward Lagrangian stochastic (bLs) model in a range of atmospheric conditions. This study's EC attempt was unsuccessful due to data collection and instrumentation issues, resulting in invalid results characterized by underestimated emissions, large negative fluxes, and cospectra/ogives that deviated from their ideal shapes. Consequently, the EC results could not be compared with the GPIM and bLS model. The bLs model demonstrated the highest accuracy for single-release single-point emissions, though it exhibited greater uncertainty than GPIM under multi-release conditions. Across the GPIM and bLs model, the most reliable quantification was achieved with 15 min averaging and a narrow 5° wind sector range. Although EC was limited in this context, future studies should consider employing a standard EC system and further optimizing GPIM and bLs approaches – particularly for complex multi-source scenarios – to enhance quantification accuracy and reduce uncertainty.

03 NATURAL GAS↗

Considering value of information when using CFD in design

This thesis presents an approach to find lower resolution CFD models that can accurately lead a designer to a correct decision at a lower computational cost. High-fidelity CFD models often contain too much information and come at a higher computational cost, limiting the designs a designer can test and how much optimization can be performed on the design. Lower model resolution is commonly used to reduce computational time. However there are no clear guidelines on how much model accuracy is required. Instead experience and intuition are used to select an appropriate lower resolution model. This thesis presents an alternative to this ad hoc method by considering the added value of the addition information provided by increasing accurate and more computationally expensive models.

99 GENERAL AND MISCELLANEOUS↗

Final Report for "Non-Accelerator Physics – Research in High Energy Physics: Dark Energy Research on DES"

One of the greatest mysteries in modern cosmology is the fact that the expansion of the universe is observed to be accelerating. This acceleration may stem from dark energy, an additional energy component of the universe, or may indicate that the theory of general relativity is incomplete on cosmological scales. The growth rate of large-scale structure in the universe and particularly the largest collapsed structures, clusters of galaxies, is highly sensitive to the underlying cosmology. Clusters will provide one of the single most precise methods of constraining dark energy with the ongoing Dark Energy Survey (DES). The accuracy of the cosmological constraints derived from DES clusters necessarily depends on having an optimized and well-calibrated algorithm for selecting clusters as well as an optical richness estimator whose mean relation and scatter compared to cluster mass are precisely known. Calibrating the galaxy cluster richness-mass relation and its scatter was the focus of the funded work. Specifically, we employ X-ray observations and optical spectroscopy with the Keck telescopes of optically-selected clusters to calibrate the relationship between optical richness (the number of galaxies in a cluster) and underlying mass. This work also probes aspects of cluster selection like the accuracy of cluster centering which are critical to weak lensing cluster studies.

43 PARTICLE ACCELERATORS↗

Evaluation of similarity measures for use in the intensity-based rigid 2D-3D registration for patient positioning in radiotherapy

Purpose: Rigid 2D-3D registration is an alternative to 3D-3D registration for cases where largely bony anatomy can be used for patient positioning in external beam radiation therapy. In this article, the authors evaluated seven similarity measures for use in the intensity-based rigid 2D-3D registration using a variation in Skerl's similarity measure evaluation protocol. Methods: The seven similarity measures are partitioned intensity uniformity, normalized mutual information (NMI), normalized cross correlation (NCC), entropy of the difference image, pattern intensity (PI), gradient correlation (GC), and gradient difference (GD). In contrast to traditional evaluation methods that rely on visual inspection or registration outcomes, the similarity measure evaluation protocol probes the transform parameter space and computes a number of similarity measure properties, which is objective and optimization method independent. The variation in protocol offers an improved property in the quantification of the capture range. The authors used this protocol to investigate the effects of the downsampling ratio, the region of interest, and the method of the digitally reconstructed radiograph (DRR) calculation [i.e., the incremental ray-tracing method implemented on a central processing unit (CPU) or the 3D texture rendering method implemented on a graphics processing unit (GPU)] on the performance of the similarity measures. The studies were carried out using both the kilovoltage (kV) and the megavoltage (MV) images of an anthropomorphic cranial phantom and the MV images of a head-and-neck cancer patient. Results: Both the phantom and the patient studies showed the 2D-3D registration using the GPU-based DRR calculation yielded better robustness, while providing similar accuracy compared to the CPU-based calculation. The phantom study using kV imaging suggested that NCC has the best accuracy and robustness, but its slow function value change near the global maximum requires a stricter termination condition for an optimization method. The phantom study using MV imaging indicated that PI, GD, and GC have the best accuracy, while NCC and NMI have the best robustness. The clinical study using MV imaging showed that NCC and NMI have the best robustness. Conclusions: The authors evaluated the performance of seven similarity measures for use in 2D-3D image registration using the variation in Skerl's similarity measure evaluation protocol. The generalized methodology can be used to select the best similarity measures, determine the optimal or near optimal choice of parameter, and choose the appropriate registration strategy for the end user in his specific registration applications in medical imaging.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Three-Dimensional High-Order Spectral Finite Volume Method for Unstructured Grids

Many areas require a very high-order accurate numerical solution of conservation laws for complex shapes. This paper deals with the extension to three dimensions of the Spectral Finite Volume (SV) method for unstructured grids, which was developed to solve such problems. We first summarize the limitations of traditional methods such as finite-difference, and finite-volume for both structured and unstructured grids. We then describe the basic formulation of the spectral finite volume method. What distinguishes the SV method from conventional high-order finite-volume methods for unstructured triangular or tetrahedral grids is the data reconstruction. Instead of using a large stencil of neighboring cells to perform a high-order reconstruction, the stencil is constructed by partitioning each grid cell, called a spectral volume (SV), into 'structured' sub-cells, called control volumes (CVs). One can show that if all the SV cells are partitioned into polygonal or polyhedral CV sub-cells in a geometrically similar manner, the reconstructions for all the SVs become universal, irrespective of their shapes, sizes, orientations, or locations. It follows that the reconstruction is reduced to a weighted sum of unknowns involving just a few simple adds and multiplies, and those weights are universal and can be pre-determined once for all. The method is thus very efficient, accurate, and yet geometrically flexible. The most critical part of the SV method is the partitioning of the SV into CVs. In this paper we present the partitioning of a tetrahedral SV into polyhedral CVs with one free parameter for polynomial reconstructions up to degree of precision five. (Note that the order of accuracy of the method is one order higher than the reconstruction degree of precision.) The free parameter will be determined by minimizing the Lebesgue constant of the reconstruction matrix or similar criteria to obtain optimized partitions. The details of an efficient, parallelizable code to solve three-dimensional problems for any order of accuracy are then presented. Important aspects of the data structure are discussed. Comparisons with the Discontinuous Galerkin (DG) method are made. Numerical examples for wave propagation problems are presented.

Liu, Yen↗

Integrating deep neural networks with full-waveform inversion: Reparameterization, regularization, and uncertainty quantification

Full-waveform inversion (FWI) is an accurate imaging approach for modeling the velocity structure by minimizing the misfit between recorded and predicted seismic waveforms. However, the strong nonlinearity of FWI resulting from fitting oscillatory waveforms can trap the optimization in local minima. We have adopted a neural-network-based full-waveform inversion (NNFWI) method that integrates deep neural networks with FWI by representing the velocity model with a generative neural network. Neural networks can naturally introduce spatial correlations as regularization to the generated velocity model, which suppresses noise in the gradients and mitigates local minima. Furthermore, the velocity model generated by neural networks is input to the same partial differential equation (PDE) solvers used in conventional FWI. The gradients of the neural networks and PDEs are calculated using automatic differentiation, which back propagates gradients through the acoustic PDEs and neural network layers to update the weights of the generative neural network. Experiments on 1D velocity models, the Marmousi model, and the 2004 BP model determine that NNFWI can mitigate local minima, especially for imaging high-contrast features such as salt bodies, and it significantly improves the inversion in the presence of noise. Adding dropout layers to the neural network model also allows analyzing the uncertainty of the inversion results through Monte Carlo dropout. NNFWI opens a new pathway to combine deep learning and FWI for exploiting the characteristics of deep neural networks and the high accuracy of PDE solvers. Because NNFWI does not require extra training data and optimization loops, it provides an attractive and straightforward alternative to conventional FWI.

58 GEOSCIENCES↗

Optimal strategy for polarization modulation in the LSPE-SWIPE experiment

Context. Cosmic microwave background (CMB) B-mode experiments are required to control systematic effects with an unprecedented level of accuracy. Polarization modulation by a half wave plate (HWP) is a powerful technique able to mitigate a large number of the instrumental systematics. Aims. Our goal is to optimize the polarization modulation strategy of the upcoming LSPE-SWIPE balloon-borne experiment, devoted to the accurate measurement of CMB polarization at large angular scales. Methods. We departed from the nominal LSPE-SWIPE modulation strategy (HWP stepped every 60 s with a telescope scanning at around 12 deg/s) and performed a thorough investigation of a wide range of possible HWP schemes (either in stepped or continuously spinning mode and at different azimuth telescope scan-speeds) in the frequency, map and angular power spectrum domain. In addition, we probed the effect of high-pass and band-pass filters of the data stream and explored the HWP response in the minimal case of one detector for one operation day (critical for the single-detector calibration process). We finally tested the modulation performance against typical HWP-induced systematics. Results. Our analysis shows that some stepped HWP schemes, either slowly rotating or combined with slow telescope modulations, represent poor choices. Moreover, our results point out that the nominal configuration may not be the most convenient choice. While a large class of spinning designs provides comparable results in terms of pixel angle coverage, map-making residuals and BB power spectrum standard deviations with respect to the nominal strategy, we find that some specific configurations (e.g., a rapidly spinning HWP with a slow gondola modulation) allow a more efficient polarization recovery in more general real-case situations. Conclusions. Although our simulations are specific to the LSPE-SWIPE mission, the general outcomes of our analysis can be easily generalized to other CMB polarization experiments.

79 ASTRONOMY AND ASTROPHYSICS↗

Accurate and Accelerated Neuromorphic Network Design Leveraging A Bayesian Hyperparameter Pareto Optimization Approach

Neuromorphic systems allow for extremely efficient hardware implementations for neural networks (NNs). In recent years, several algorithms have been presented to train spiking NNs (SNNs) for neuromorphic hardware. However, SNNs often provide lower accuracy than their artificial NNs (ANNs) counterparts or require computationally expensive and slow training/inference methods. To close this gap, designers typically rely on reconfiguring SNNs through adjustments in the neuron/synapse model or training algorithm itself. Nevertheless, these steps incur significant design time, while still lacking the desired improvement in terms of training/inference times (latency). Designing SNNs that can mimic the accuracy of ANNs with reasonable training times is an exigent challenge in neuromorphic computing. In this work, we present an alternative approach that looks at such designs as an optimization problem rather than algorithm or architecture redesign. We develop a versatile multiobjective hyperparameter optimization (HPO) for automatically tuning HPs of two state-of-the-art SNN training algorithms, SLAYER and HYBRID. We emphasize that, to the best of our knowledge, this is the first work trying to improve SNNs’ computational efficiency, accuracy, and training time using an efficient HPO. We demonstrate significant performance improvements for SNNs on several datasets without the need to redesign or invent new training algorithms/architectures. Our approach results in more accurate networks with lower latency and, in turn, higher energy efficiency than previous implementations. In particular, we demonstrate improvement in accuracy and more than 5× reduction in the training/inference time for the SLAYER algorithm on the DVS Gesture dataset. In the case of HYBRID, we demonstrate 30% reduction in timesteps while surpassing the accuracy of the state-of-the-art networks on CIFAR10. Further, our analysis suggests that even a seemingly minor change in HPs could change the accuracy by 5 - 6×.

Parsa, Maryam↗

TESS Discoveries

TESS discoveries have already become fruitful targets for HST follow up, and this will undoubtedly continue through the JWST era. In addition to providing targets for atmospheric characterization, TESS' Level One Science Requirement is to measure masses for 50 transiting planets smaller than 4 Earth radii. A full suite of ground based facilities will be working together to optimize the TESS science yield. Somewhat surprisingly though, no study has quantified the accuracy of mass constraints required to yield robust atmospheric properties of small planets. Previous work showed that the mass of a transiting exoplanet could be inferred from its transmission spectrum alone. The method leverages the effect of the planet's surface gravity on the atmospheric scale height, which in turn influences the transmission spectrum. However, significant degeneracies exist between transmission spectra of planets with different masses and compositions, making difficult to unambiguously determine the planet's mass and composition in many cases. I will present the first quantitative answer to this pressing question. Our analysis places definitive limits on how accurate mass constraints need to be in order to unambiguously determine atmospheric composition for a diverse array of planets ranging from terrestrial-size (TRAPPIST-1-like) to mini-Neptunes and hot Jupiters. These results broadly impact the community of scientists working on exoplanets -- from the full breadth of ground based observers conducting TESS follow-up, to those studying planet populations, and finally to those planning atmospheric investigations. This is particularly timely as the STScI Director charged the HST-TESS Advisory committee to report to the Space Telescope Users Committee on how HST can best support follow-up observations of TESS exoplanet discoveries. The community needs to determine optimal strategies for maximizing the rapid scientific return from TESS targets.

Batalha, Natasha E.↗

Reconstructing the stellar mass distributions of galaxies using S{sup 4}G IRAC 3.6 and 4.5 μm images. II. The conversion from light to mass

We present a new approach for estimating the 3.6 μm stellar mass-to-light (M/L) ratio Y{sub 3.6} in terms of the [3.6]-[4.5] colors of old stellar populations. Our approach avoids several of the largest sources of uncertainty in existing techniques using population synthesis models. By focusing on mid-IR wavelengths, we gain a virtually dust extinction-free tracer of the old stars, avoiding the need to adopt a dust model to correctly interpret optical or optical/near-IR colors normally leveraged to assign the mass-to-light ratio Y. By calibrating a new relation between near-IR and mid-IR colors of giant stars observed in GLIMPSE we also avoid the discrepancies in model predictions for the [3.6]-[4.5] colors of old stellar populations due to uncertainties in the molecular line opacities assumed in template spectra. We find that the [3.6]-[4.5] color, which is driven primarily by metallicity, provides a tight constraint on Y{sub 3.6}, which varies intrinsically less than at optical wavelengths. The uncertainty on Y{sub 3.6} of ∼0.07 dex due to unconstrained age variations marks a significant improvement on existing techniques for estimating the stellar M/L with shorter wavelength data. A single Y{sub 3.6} = 0.6 (assuming a Chabrier initial mass function (IMF)), independent of [3.6]-[4.5] color, is also feasible because it can be applied simultaneously to old, metal-rich and young, metal-poor populations, and still with comparable (or better) accuracy (∼0.1 dex) than alternatives. We expect our Y{sub 3.6} to be optimal for mapping the stellar mass distributions in S{sup 4}G galaxies, for which we have developed an independent component analysis technique to first isolate the old stellar light at 3.6 μm from nonstellar emission (e.g., hot dust and the 3.3 polycyclic aromatic hydrocarbon feature). Our estimate can also be used to determine the fractional contribution of nonstellar emission to global (rest-frame) 3.6 μm fluxes, e.g., in WISE imaging, and establishes a reliable basis for exploring variations in the stellar IMF.

79 ASTRONOMY AND ASTROPHYSICS↗

An Optimization-Based Coupling of Reduced Order Models with an Efficient Reduced Adjoint Basis Generation Approach

Optimization-based coupling (OBC) is an attractive alternative to traditional Lagrange multiplier approaches in multiple modeling and simulation contexts. However, application of OBC to time-dependent problems has been hindered by the computational cost of finding the stationary points of the associated Lagrangian, which requires primal and adjoint solves. This issue can be mitigated by using OBC in conjunction with computationally efficient reduced order models (ROMs). To demonstrate the potential of this combination, in this paper, we develop an optimization-based ROM-ROM coupling for a transient advection-diffusion transmission problem. We pursue the “optimize-then-reduce” path toward solving the minimization problem at each time step and solve reduced space adjoint system of equations, where the main challenge in this formulation is the generation of adjoint snapshots and reduced bases for the adjoint systems required by the optimizer. One of the main contributions of the paper is a new technique for an efficient adjoint snapshot collection for gradient-based optimizers in the context of optimization-based ROM-ROM couplings. In conclusion, we present numerical studies demonstrating the accuracy of the approach along with comparison between various approaches for selecting a reduced order basis for the adjoint systems, including decay of snapshot energy, average iteration counts, and timings.

coupled problems↗

Integrated structure-control optimization of space structures

In the past few years there has been substantial interest in the integrated design of large space structures and their vibration control systems. This paper overviews two aspects of integrated control-structure design. The first is the calculation of derivatives of control system stability margins with respect to structural parameters, and the effect of modal truncation on the accuracy of these derivatives. The second aspect is the various formulations used to define an integrated design optimization problem.

Haftka, Raphael T.↗

Response of joint dominated space structures

An approximate method is developed for estimating the transient response of nonlinear systems in terms of linearized modes of response. Its advantages are that it is computationally more efficient than the time integration method and that it is possible to view the design problem in the more traditional physical terms of modal response. The major drawback of the approximate method is loss of accuracy. It seems that both approximate methods and time integration have their roles in design. Approximate methods provide efficient tools for performing parametric studies and they supply physical insights into how to optimize system performance that are not easily inferred from strictly numerical methods. Time integration provides a method for assessing the accuracy of the approximate solution for key simulations and for fine tuning the final design. In the procedure presented the nonlinear system is approximated by an equivalent linear system in which the system parameters are constant over the range of transient response.

Source record↗

High-Order Spectral Volume Method for the Navier-Stokes Equations On Unstructured Tetrahedral Grids (Final Report)

The overriding objective for this project is to develop an efficient and accurate method for capturing strong discontinuities and fine smooth flow structures of disparate length scales with unstructured grids, and demonstrate its potentials for problems relevant to DOE. More specifically, we plan to achieve the following objectives: 1. Extend the SV method to three dimensions, and develop a fourth-order accurate SV scheme for tetrahedral grids. Optimize the SV partition by minimizing a form of the Lebesgue constant. Verify the order of accuracy using the scalar conservation laws with an analytical solution; 2. Extend the SV method to Navier-Stokes equations for the simulation of viscous flow problems. Two promising approaches to compute the viscous fluxes will be tested and analyzed; 3. Parallelize the 3D viscous SV flow solver using domain decomposition and message passing. Optimize the cache performance of the flow solver by designing data structures minimizing data access times; 4. Demonstrate the SV method with a wide range of flow problems including both discontinuities and complex smooth structures. The objectives remain the same as those outlines in the original proposal. We anticipate no technical obstacles in meeting these objectives.

97 MATHEMATICS AND COMPUTING↗

An Efficient Referencing And Sample Positioning System To Investigate Heterogeneous Substances With Combined Microfocused Synchrotron X-ray Techniques

A referencing and sample positioning system has been developed to transfer object positions measured with an offline microscope to a synchrotron experimental station. The accuracy should be sufficient to deal with heterogeneous samples on micrometer scale. Together with an online fluorescence mapping visualisation the optical alignment helps to optimize measuring procedures for combined microfocused X-ray techniques.

39↗