Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Kriging”

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 145 records · Page 8

Tuning a RANS k-e model for jet-in-crossflow simulations.

We develop a novel calibration approach to address the problem of predictive k e RANS simulations of jet-incrossflow. Our approach is based on the hypothesis that predictive k e parameters can be obtained by estimating them from a strongly vortical flow, specifically, flow over a square cylinder. In this study, we estimate three k e parameters, C%CE%BC, Ce2 and Ce1 by fitting 2D RANS simulations to experimental data. We use polynomial surrogates of 2D RANS for this purpose. We conduct an ensemble of 2D RANS runs using samples of (C%CE%BC;Ce2;Ce1) and regress Reynolds stresses to the samples using a simple polynomial. We then use this surrogate of the 2D RANS model to infer a joint distribution for the k e parameters by solving a Bayesian inverse problem, conditioned on the experimental data. The calibrated (C%CE%BC;Ce2;Ce1) distribution is used to seed an ensemble of 3D jet-in-crossflow simulations. We compare the ensemble's predictions of the flowfield, at two planes, to PIV measurements and estimate the predictive skill of the calibrated 3D RANS model. We also compare it against 3D RANS predictions using the nominal (uncalibrated) values of (C%CE%BC;Ce2;Ce1), and find that calibration delivers a significant improvement to the predictive skill of the 3D RANS model. We repeat the calibration using surrogate models based on kriging and find that the calibration, based on these more accurate models, is not much better that those obtained with simple polynomial surrogates. We discuss the reasons for this rather surprising outcome.

Lefantzi, Sophia↗

Design of Experiments, Model Calibration and Data Assimilation

This presentation provides an overview of emulation, calibration and experiment design for computer experiments. Emulation refers to building a statistical surrogate from a carefully selected and limited set of model runs to predict unsampled outputs. The standard kriging approach to emulation of complex computer models is presented. Calibration refers to the process of probabilistically constraining uncertain physics/engineering model inputs to be consistent with observed experimental data. An initial probability distribution for these parameters is updated using the experimental information. Markov chain Monte Carlo (MCMC) algorithms are often used to sample the calibrated parameter distribution. Several MCMC algorithms commonly employed in practice are presented, along with a popular diagnostic for evaluating chain behavior. Space-filling approaches to experiment design for selecting model runs to build effective emulators are discussed, including Latin Hypercube Design and extensions based on orthogonal array skeleton designs and imposed symmetry requirements. Optimization criteria that further enforce space-filling, possibly in projections of the input space, are mentioned. Designs to screen for important input variations are summarized and used for variable selection in a nuclear fuels performance application. This is followed by illustration of sequential experiment design strategies for optimization, global prediction, and rare event inference.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Simplified predictive models for CO 2 sequestration performance assessment

CO2 sequestration in deep saline formations is increasingly being considered as a viable strategy for the mitigation of greenhouse gas emissions from anthropogenic sources. In this context, detailed numerical simulation based models are routinely used to understand key processes and parameters affecting pressure propagation and buoyant plume migration following CO2 injection into the subsurface. As these models are data and computation intensive, the development of computationally-efficient alternatives to conventional numerical simulators has become an active area of research. Such simplified models can be valuable assets during preliminary CO2 injection project screening, serve as a key element of probabilistic system assessment modeling tools, and assist regulators in quickly evaluating geological storage projects. We present three strategies for the development and validation of simplified modeling approaches for CO2 sequestration in deep saline formations: (1) simplified physics-based modeling, (2) statisticallearning based modeling, and (3) reduced-order method based modeling. In the first category, a set of full-physics compositional simulations is used to develop correlations for dimensionless injectivity as a function of the slope of the CO2 fractional-flow curve, variance of layer permeability values, and the nature of vertical permeability arrangement. The same variables, along with a modified gravity number, can be used to develop a correlation for the total storage efficiency within the CO2 plume footprint. Furthermore, the dimensionless average pressure buildup after the onset of boundary effects can be correlated to dimensionless time, CO2 plume footprint, and storativity contrast between the reservoir and caprock. In the second category, statistical “proxy models” are developed using the simulation domain described previously with two approaches: (a) classical Box-Behnken experimental design with a quadratic response surface, and (b) maximin Latin Hypercube sampling (LHS) based design with a multidimensional kriging metamodel fit. For roughly the same number of simulations, the LHS-based metamodel yields a more robust predictive model, as verified by a k-fold cross-validation approach (with data split into training and test sets) as well by validation with an independent dataset. In the third category, a reduced-order modeling procedure is utilized that combines proper orthogonal decomposition (POD) for reducing problem dimensionality with trajectory-piecewise linearization (TPWL) in order to represent system response at new control settings from a limited number of training runs. Significant savings in computational time are observed with reasonable accuracy from the PODTPWL reduced-order model for both vertical and horizontal well problems – which could be important in the context of history matching, uncertainty quantification and optimization problems. The simplified physics and statistical learning based models are also validated using an uncertainty analysis framework. Reference cumulative distribution functions of key model outcomes (i.e., plume radius and reservoir pressure buildup) generated using a 97-run full-physics simulation are successfully validated against the CDF from 10,000 sample probabilistic simulations using the simplified models. The main contribution of this research project is the development and validation of a portfolio of simplified modeling approaches that will enable rapid feasibility and risk assessment for CO2 sequestration in deep saline formations.

54 ENVIRONMENTAL SCIENCES↗

Play Fairway Analysis: Structurally Controlled Geothermal Systems in the Eastern Great Basin Extensional Regime, Utah

A research team with membership from the University of Utah/Energy & Geoscience Institute, the University of Utah/Dept. of Geology & Geophysics, and the Utah Geological Survey, undertook a play fairway analysis (PFA) for geothermal resources in the Eastern Great Basin (EGB) extensional tectonic regime of western Utah. This is a high-priority region for geothermal exploration because active Basin and Range (B&R) extension with volcanism having a N-S strike is superimposed upon pre-existing E-W belts of plutonic rocks and large-scale structural lineaments. Cumulative heat flow along the N-S strike of the state totals approximately 5 GWt above background stable interior. Three electricity producing power plants currently exist with substantial potential for increase. Succinctly, our PFA approach aims to resolve potential sources of heat and permeability in the region, which are the two principal criteria for establishing a geothermal resource. An initial Phase 1 was carried out using only existing geoscientific data in the area. Criteria selected for focusing heat potential include direct heat flow measurements in boreholes, magnetotelluric (MT) low resistivity anomalies, fluid/gas geochemistry, and proximity to recent volcanic eruptions. Permeability is established through geological structures (fault density, critically stressed areas, seismicity, gravity), and MT low resistivity anomalies. In Phase II of this PFA project, additional geological, geophysical and geochemical data were acquired and analysis carried out primarily over promising composite common risk (CCR) areas initially identified in Phase I in order to focus prospectivity and prepare for drilling recommendations. These prospects are near the Twin Peaks rhyolite field, high heat flow areas north of the producing Cove Fort system, and geophysical structure beneath the Crater Knoll area off the northeast flank of the Mineral Mountains. Additional data included MT site fill-in, structural mapping and analysis using high-resolution imagery, gravity and on-ground mapping, Nodal 3C passive seismic collection, and passive 3He surveying. Heat source and permeability potential are again expressed in terms of their individual common risk segment (CRS) maps, with a color scheme using green for most favorable (low risk) and red for least favorable (high risk). Diverse data types are united through the technique of probability kriging, which establishes prospectivity thresholds for each data type and then computes probability of exceeding that threshold over the PFA area. Modified CRS and CCRS maps are compared to those of Phase I to highlight tighter prospectivity focus. In doing so, the prospectivity threshold for heat was increased significantly to narrow the targeting. In the final Phase III of this project where a recommended deep thermal gradient hole was sited, additional geophysical, geochemical and geological field collection and analysis was carried out to refine drill hole targeting. This includes prospect-scale MT, gravity, structure, passive seismic deployment (Cove Fort area), and detailed 3He isotope profiling. It was the recommendation of the DOE Technical Monitoring Team (TMT) that one or more holes be sited in the north-ern Cove Fort area where legacy TG gradient holes showed high cumulative heat flow. These were to be of moderate depth, 2000-3000 feet, to reach the geothermal fluid table expected to start in excess of 1000 feet depth. The drilling organization stipulated by the DOE/GTO was that of the USGS Research Drilling Program (RDP) centered in Las Vegas, NV. A detailed well plan, appended to this report, was developed principally by Dr. Ben Barker consulting to University of Utah, Dr. Steve Pye on the DOE TMT, Mr. Steven Crawford of the USGS-RDP, and the project PI Phil Wannamaker. However, temperature and possible H 2 S at the systems lead to cancellation of the drilling last-minute as this appeared outside the experience base of the USGS-RDP. We hope to have the opportunity to revisit the test drilling and expand the Play Fairway Analysis of this region at some point in the future.

15 GEOTHERMAL ENERGY↗

Verification and Validation Activities for the Multi-Fidelity Toolkit

The Multi-Fidelity Toolkit (MFTK) is a simulation tool being developed at Sandia National Laboratories for aerodynamic predictions of compressible flows over a range of physics fidelities and computational speeds. These models include the Reynolds-Averaged-Navier-Stokes (RANS) equations, the Euler equations, and modified Newtonian aerodynamics (MNA) equations, and they can be invoked independently or coupled with hierarchical Kriging to interpolate between high-fidelity simulations using lower-fidelity data. However, as with any new simulation capability, verification and validation are necessary to gather credibility evidence. This work describes formal code- and solution-verification activities as well as model validation with uncertainty considerations. Code verification is performed on the MNA model by comparing with an analytical solution for flat-plate and inclined-plate geometries. Solution-verification activities include grid-refinement studies of HIFiRE-1 wind tunnel measurements, which are used for validation, for all model fidelities. A thorough treatment of the validation comparison with prediction error and validation uncertainty is also presented.

42 ENGINEERING↗

Expanded verification and validation studies of hypersonic aerodynamics with multiple physics-fidelity models

Hypersonic aerothermodynamics is an important domain of modern multiphysics simulation. The Multi-Fidelity Toolkit is a simulation tool being developed at Sandia National Laboratories to predict aerodynamic properties for compressible flows from a range of physics fidelities and computational speeds. These models include the Reynolds-averaged Navier–Stokes (RANS) equations, the Euler equations with momentum-energy integral technique (MEIT), and modified Newtonian aerodynamics with flat-plate boundary layer (MNA+FPBL) equations, and they can be invoked independently or coupled with hierarchical Kriging to interpolate between high-fidelity simulations using lower-fidelity data. However, as with any new simulation capability, verification and validation are necessary to gather credibility evidence. This work describes formal code- and solution-verification activities, as well as model validation with uncertainty considerations. Code verification activities on the MNA+FPBL model build on previous work by focusing on the viscous portion of the model. Viscous quantities of interest are compared against those from an analytical solution for flat-plate, inclined-plate, and cone geometries. The code verification methodology for the MEIT model is also presented. Test setup and results of code verification tests on the laminar and turbulent models within MEIT are shown. Solution-verification activities include grid-refinement studies on simulations that model the HIFiRE-1 wind tunnel experiments. These experiments are used for validation of all model fidelities. A thorough validation comparison with prediction error and uncertainty is also presented. Three additional HIFiRE-1 experimental runs are simulated in this study, and the solution verification and validation work examines the effects of the associated parameter changes on model performance. Finally, a study is presented that compares the computational costs and fidelities from each of the different models.

42 ENGINEERING↗

Integration of Soft Data Into Geostatistical Simulation of Categorical Variables

Uncertain or indirect “soft” data, such as geologic interpretation, driller’s logs, geophysical logs or imaging, offer potential constraints or “soft conditioning” to stochastic models of discrete categorical subsurface variables in hydrogeology such as hydrofacies. Previous bivariate geostatistical simulation algorithms have not fully addressed the impact of data uncertainty in formulation of the (co) kriging equations and the objective function in simulated annealing (or quenching). This paper introduces the geostatistical simulation code tsim-s, which accounts for categorical data uncertainty through a data “hardness” parameter. In generating geostatistical realizations with tsim-s, the uncertainty inherent to soft conditioning is factored into both 1) the data declustering and spatial correlation functions in cokriging and 2) the acceptance probability for change of category in simulated quenching. The degree or sensitivity to which soft data conditions a realization as a function of hardness can be quantified by mapping category probabilities derived from multiple realizations. In addition to point or borehole data, arrays of data (e.g., as derived from a depth-dependency function, probability map, or “prior realization”) can be used as soft conditioning. The tsim-s algorithm provides a theoretically sound and general framework for integrating datasets of variable location, resolution, and uncertainty into geostatistical simulation of categorical variables. A practical example shows how tsim-s is capable of generating a large-scale three-dimensional simulation including curvilinear features.

54 ENVIRONMENTAL SCIENCES↗

Towards Hyper-Dimensional Variography Using the Product-Sum Covariance Model

Modeling hyper-dimensional spatial variability is a complex task from both practical and theoretical standpoints. In this paper we develop a method for modeling hyper-dimensional covariance (variogram) structures using the product-sum covariance model initially developed to model spatio-temporal variability. We show that the product-sum model can be used recursively up to an arbitrarily large number of dimensions while preserving relative modeling simplicity and yielding valid covariance models. The method can be used to model variability in anisotropic conditions with multiple axes of anisotropy or when temporal evolution is involved, and thus is applicable to “full anisotropic 3D+time” situations often encountered in environmental sciences. It requires fewer assumptions than the traditional product-sum modeling approach. The new method also presents an alternative to classical approaches to modeling zonal anisotropy and requires fewer parameters to be estimated from data. We present an example by applying the method in conjunction with ordinary kriging to map photosynthetically-active radiation (PAR) for 2006, in Oklahoma, CA, USA and to explore effects of spatio-temporal variability in PAR on gross primary productivity (GPP).

54 ENVIRONMENTAL SCIENCES↗

Refractivity Observations from Radar Phase Measurements: The 22 May 2002 Dryline Case during IHOP Project

The dryline, often associated with the development of severe storms in the Southern Great Plains of the United States of America, is a boundary layer phenomenon that occurs when a warm and moist air mass from the Gulf of Mexico meets a hot and dry air mass from the southwest desert area. An accurate knowledge of the water vapor spatio-temporal variability in the lower part of the atmosphere is crucial for a better understanding of the evolution of the dryline. The tropospheric refractivity, directly related to water vapor content, is a proxy for the water vapor content of the troposphere. It has already been demonstrated that the refractivity and the refractivity vertical gradient can be jointly estimated from radar phase measurements. In fact, it has been shown that using kriging interpolation techniques, accurate refractivity maps within the coverage area of the radar can be obtained with high temporal resolution. In this paper, a detailed analysis of the time series of radar-based refractivity maps obtained during a dryline that occurred on the afternoon of 22 May 2002 during the International H 2 O Project is presented. Comparisons between the time series of radar refractivity maps, obtained with the NCAR S-Pol radar, and the refractivity measurements derived from automatic ground-based weather stations and the AERI instrument, placed at different locations within the coverage area of the NCAR S-Pol radar, demonstrate the accuracy of radar refractivity estimates even for highly variable conditions, both in time and space, in the troposphere. Correlation coefficients higher than 0.95 are obtained in all weather station locations. Regarding the RMSE, errors less than 6 N-units are obtained for all cases, being even as low as 2.92 N-units at some locations.

54 ENVIRONMENTAL SCIENCES↗

Monitoring global snow cover

A snow model that supports the daily, operational analysis of global snow depth and age has been developed. It provides improved spatial interpolation of surface reports by incorporating digital elevation data, and by the application of regionalized variables (kriging) through the use of a global snow depth climatology. Where surface observations are inadequate, the model applies satellite remote sensing. Techniques for extrapolation into data-void mountain areas and a procedure to compute snow melt are also contained in the model.

Armstrong, Richard↗

Cokriging with ground-based radiometry

The formulas for cokriging and a coherent coregionalization model are presented. The model is applied to design sampling strategies for surveys using a ground-based radiometer. Results indicate that cokriging based on measured radiation is nine times as efficient as kriging the cover alone. It is concluded that cokriging in conjunction with ground-based radiometry provides an economical and operational technique for using reflectance to estimate the earth surface properties.

Atkinson, P. M.↗

Design and analysis of numerical experiments

Methods for the design and analysis of numerical experiments that are especially useful and efficient in multidimensional parameter spaces are presented. The analysis method, which is similar to kriging in the spatial analysis literature, fits a statistical model to the output of the numerical model. The method is applied to a fully nonlinear, global, equivalent-barotropic dynamical model. The statistical model also provides estimates for the uncertainty of predicted numerical model output, which can provide guidance on where in the parameter space to conduct further experiments, if necessary. The method can provide significant improvements in the efficiency with which numerical sensitivity experiments are conducted.

Bowman, Kenneth P.↗

Validating spatial structure in canopy water content using geostatistics

Heterogeneity in ecological phenomena are scale dependent and affect the hierarchical structure of image data. AVIRIS pixels average reflectance produced by complex absorption and scattering interactions between biogeochemical composition, canopy architecture, view and illumination angles, species distributions, and plant cover as well as other factors. These scales affect validation of pixel reflectance, typically performed by relating pixel spectra to ground measurements acquired at scales of 1m(exp 2) or less (e.g., field spectra, foilage and soil samples, etc.). As image analysis becomes more sophisticated, such as those for detection of canopy chemistry, better validation becomes a critical problem. This paper presents a methodology for bridging between point measurements and pixels using geostatistics. Geostatistics have been extensively used in geological or hydrogeolocial studies but have received little application in ecological studies. The key criteria for kriging estimation is that the phenomena varies in space and that an underlying controlling process produces spatial correlation between the measured data points. Ecological variation meets this requirement because communities vary along environmental gradients like soil moisture, nutrient availability, or topography.

Sanderson, E. W.↗

Numerical Optimization Using Computer Experiments

Engineering design optimization often gives rise to problems in which expensive objective functions are minimized by derivative-free methods. We propose a method for solving such problems that synthesizes ideas from the numerical optimization and computer experiment literatures. Our approach relies on kriging known function values to construct a sequence of surrogate models of the objective function that are used to guide a grid search for a minimizer. Results from numerical experiments on a standard test problem are presented.

Trosset, Michael W.↗

Biweekly Maps of Wind Stress for the North Pacific from the ERS-1 Scatterometer

The European Remote-sensing Satellite (ERS-1) was launched in July 1991 and contained several instruments for observing the Earth's ocean including a wind scatterometer. The scatterometer measurements were processed by the European Space Agency (ESA) and the Jet Propulsion Laboratory (JPL). JPL reprocessed (Freilich and Dunbar, 1992) the ERS-1 backscatter measurements to produced a 'value added' data set that contained the ESA wind vector as well as a set of up to four ambiguities. These ambiguities were further processed using a maximum-likelihood estimation (MLE) and a median filter to produce a 'selected vector.' This report describes a technique developed to produce time-averaged wind field estimates with their expected errors using only scatterometer wind vectors. The processing described in this report involved extracting regions of interest from the data tapes, checking the quality and creating the wind field estimate. This analysis also includes the derivation of biweekly average wind vectors over the North Pacific Ocean at a resolution of 0.50 x 0.50. This was done with an optimal average algorithm temporally and an over-determined biharmonic spline spatially. There have been other attempts at creating gridded wind files from ERS-1 winds, e.g., kriging techniques (Bentamy et al., 1996) and successive corrections schemes (Tang and Liu, 1996). There are several inherent problems with the ERS-1 scatterometer. Since this is a multidisciplinary mission, the satellite is flown in different orbits optimized for each phase of the mission. The scatterometer also shares several sub-systems with the Synthetic Aperture Radar (SAR) and cannot be operated while the SAR is in operation. The scatterometer is also a single-sided instrument and only measures backscatter along the right side of the satellite. The processing described here generates biweekly wind maps during the wktwo years analysis period regardless of the satellite orbit or missing data.

SYNTHETIC APERTURE RADAR↗

Metamodels for Computer-Based Engineering Design: Survey and Recommendations

The use of statistical techniques to build approximations of expensive computer analysis codes pervades much of todays engineering design. These statistical approximations, or metamodels, are used to replace the actual expensive computer analyses, facilitating multidisciplinary, multiobjective optimization and concept exploration. In this paper we review several of these techniques including design of experiments, response surface methodology, Taguchi methods, neural networks, inductive learning, and kriging. We survey their existing application in engineering design and then address the dangers of applying traditional statistical techniques to approximate deterministic computer analysis codes. We conclude with recommendations for the appropriate use of statistical approximation techniques in given situations and how common pitfalls can be avoided.

Simpson, Timothy W.↗

On the Use of Statistics in Design and the Implications for Deterministic Computer Experiments

Perhaps the most prevalent use of statistics in engineering design is through Taguchi's parameter and robust design -- using orthogonal arrays to compute signal-to-noise ratios in a process of design improvement. In our view, however, there is an equally exciting use of statistics in design that could become just as prevalent: it is the concept of metamodeling whereby statistical models are built to approximate detailed computer analysis codes. Although computers continue to get faster, analysis codes always seem to keep pace so that their computational time remains non-trivial. Through metamodeling, approximations of these codes are built that are orders of magnitude cheaper to run. These metamodels can then be linked to optimization routines for fast analysis, or they can serve as a bridge for integrating analysis codes across different domains. In this paper we first review metamodeling techniques that encompass design of experiments, response surface methodology, Taguchi methods, neural networks, inductive learning, and kriging. We discuss their existing applications in engineering design and then address the dangers of applying traditional statistical techniques to approximate deterministic computer analysis codes. We conclude with recommendations for the appropriate use of metamodeling techniques in given situations and how common pitfalls can be avoided.

Simpson, Timothy W.↗

Construction of Response Surface with Higher Order Continuity and Its Application to Reliability Engineering

The usefulness of piecewise polynomials with C1 and C2 derivative continuity for response surface construction method is examined. A Moving Least Squares (MLS) method is developed and compared with four other interpolation methods, including kriging. First the selected methods are applied and compared with one another in a two-design variables problem with a known theoretical response function. Next the methods are tested in a four-design variables problem from a reliability-based design application. In general the piecewise polynomial with higher order derivative continuity methods produce less error in the response prediction. The MLS method was found to be superior for response surface construction among the methods evaluated.

Krishnamurthy, T.↗