Search NASA⌕ Search

Engineering topics

Kamath, Chandrika

Publications and source records attributed to Kamath, Chandrika.

At least 19 records

Spatio-Temporal Surrogates for Interaction of a Jet with High Explosives: Part I - Analysis with a Small Sample Size

Computer simulations, especially of complex phenomena, can be expensive, requiring high-performance computing resources. Often, to understand a phenomenon, multiple simulations are run, each with a different set of simulation input parameters. These data are then used to create an interpolant, or surrogate, relating the simulation outputs to the corresponding inputs. When the inputs and outputs are scalars, a simple machine learning model can suffice. However, when the simulation outputs are vector valued, available at locations in two or three spatial dimensions, often with a temporal component, creating a surrogate is more challenging. In this report, we use a two-dimensional problem of a jet interacting with high explosives to understand how we can build high-quality surrogates. The characteristics of our data set are unique - the vector-valued outputs from each simulation are available at over two million spatial locations; each simulation is run for a relatively small number of time steps; the size of the computational domain varies with each simulation; and resource constraints limit the number of simulations we can run. We show how we analyze these extremely large data-sets, set the parameters for the algorithms used in the analysis, and use simple ways to improve the accuracy of the spatio-temporal surrogates without substantially increasing the number of simulations required.

97 MATHEMATICS AND COMPUTING↗

Spatio-Temporal Surrogates for Interaction of a Jet with High Explosives: Part II - Clustering Extremely High-Dimensional Grid-Based Data

Building an accurate surrogate model for the spatio-temporal outputs of a computer simulation is a challenging task. A simple approach to improve the accuracy of the surrogate is to cluster the outputs based on similarity and build a separate surrogate model for each cluster. This clustering is relatively straightforward when the output at each time step is of moderate size. However, when the spatial domain is represented by a large number of grid points, numbering in the millions, the clustering of the data becomes more challenging. In this report, we consider output data from simulations of a jet interacting with high explosives. These data are available on spatial domains of different sizes, at grid points that vary in their spatial coordinates, and in a format that distributes the output across multiple files at each time step of the simulation. We first describe how we bring these data into a consistent format prior to clustering. Borrowing the idea of random projections from data mining, we reduce the dimension of our data by a factor of thousand, making it possible to use the iterative k-means method for clustering. We show how we can use the randomness of both the random projections, and the choice of initial centroids in k-means clustering, to determine the number of clusters in our data set. Our approach makes clustering of extremely high dimensional data tractable, generating meaningful cluster assignments for our problem, despite the approximation introduced in the random projections.

97 MATHEMATICS AND COMPUTING↗

Classification of Orbits in Poincare Maps Using Machine Learning

Poincare plots, also called Poincare maps, are used by plasma physicists to understand the behavior of magnetically confined plasma in numerical simulations of a tokamak. These plots are created by the intersection of field lines with a two-dimensional poloidal plane that is perpendicular to the axis of the torus representing the tokamak. A plot is composed of multiple orbits, each created by a different field line as it goes around the torus. Each orbit can have one of four distinct shapes, or classes, that indicate changes in the topology of the magnetic fields confining the plasma. Given the (x, y) coordinates of the points that form an orbit, the analysis task is to assign a class to the orbit, a task that appears ideally suited for a machine learning approach. In this paper, we describe how we overcame two major challenges in solving this problem - creating a high-quality training set, with few mislabeled orbits, and converting the coordinates of the points into features that are discriminating, despite the variation within the orbits of a class and the apparent similarities between orbits of different classes. Our automated approach is not only more objective and accurate than visual classification, but is also less tedious, making it easier for plasma physicists to analyze the topology of magnetic fields from numerical simulations of the tokamak.

97 MATHEMATICS AND COMPUTING↗

Intelligent Sampling for Surrogate Modeling, Hyperparameter Optimization, and Data Analysis

Sampling techniques are used in many fields, including design of experiments, image processing, and graphics. The techniques in each field are designed to meet the constraints specific to that field such as uniform coverage of the range of each dimension or random samples that are at least a certain distance apart from each other. When an application imposes new constraints, for example, by requiring samples in a non-rectangular domain or the addition of new samples to an existing set, a common solution is to modify the algorithm currently in use, often with less than satisfactory results. As an alternative, we propose the concept of intelligent sampling, where we devise algorithms specifically tailored to meet our sampling needs, either by creating new algorithms or by modifying suitable algorithms from other fields. Surprisingly, both qualitative and quantitative comparisons indicate that some relatively simple algorithms can be easily modified to meet the many sampling requirements of surrogate modeling, hyperparameter optimization, and data analysis; these algorithms outperform their more sophisticated counterparts currently in use, resulting in better use of time and computer resources.

97 MATHEMATICS AND COMPUTING↗

Randomized Algorithms for Scientific Computing (RASC)

Randomized algorithms have propelled advances in artificial intelligence (AI) and represent a foundational research area in advancing AI for Science. Future advancements in DOE Office of Science priority areas such as climate science, astrophysics, fusion, advanced materials, combustion, and quantum computing all require randomized algorithms for surmounting challenges of complexity, robustness, and scalability. Advances in data collection and numerical simulation have changed the dynamics of scientific research and motivate the need for randomized algorithms. For instance, advances in imaging technologies such as X-ray ptychography, electron microscopy, electron energy loss spectroscopy, or adaptive optics lattice light-sheet microscopy collect hyperspectral imaging and scattering data in terabytes, at breakneck speed enabled by state-of-the-art detectors. The data collection is exceptionally fast compared with its analysis. Likewise, advances in high-performance architectures have made exascale computing a reality and changed the economies of scientific computing in the process. Floating-point operations that create data are essentially free in comparison with data movement. Thus far, most approaches have focused on creating faster hardware. Ironically, this faster hardware has exacerbated the problem by making data still easier to create. Under such an onslaught, scientists often resort to heuristic deterministic sampling schemes (e.g., low-precision arithmetic, sampling every nth element) and sacrifice potentially valuable accuracy. Dramatically better results can be achieved via randomized algorithms, reducing the data size as much as or more than naive deterministic subsampling can achieve, while retaining the high accuracy of computing on the full data set. By randomized algorithms we mean those algorithms that employ some form of randomness in internal algorithmic decisions to accelerate time to solution, increase scalability, or improve reliability. Examples include matrix sketching for solving large-scale least-squares problems (see Figure 1) and stochastic gradient descent for training machine learning models. We are not recommending heuristic methods but rather randomized algorithms that have certificates of correctness and probabilistic guarantees of optimality and near-optimality. Such approaches can be useful beyond acceleration, for example, in understanding how to avoid measure zero worst-case scenarios that plague methods such as QR matrix factorization.

97 MATHEMATICS AND COMPUTING↗

Data Mining for Faster, Interpretable Solutions to Inverse Problems:A Case Study Using Additive Manufacturing

Solving inverse problems, where we nd the input values that result in desired values of outputs, can be challenging. The solution process is often computationally expensive and it can be di cult to interpret the solution in high-dimensional input spaces. In this paper, we use a problem from additive manufacturing to address these two issues with the intent of making it easier to solve inverse problems and exploit their results. First, focusing on Gaussian process surrogates that are used to solve inverse problems, we describe how a simple modi cation to the idea of tapering can substantially speed up the surrogate without losing accuracy in prediction. Second, we demonstrate that Kohonen self-organizing maps can be used to visualize and interpret the solution to the inverse problem in the high-dimensional input space. For our data set, as not all input dimensions are equally important, we show that using weighted distances results in a better organized map that makes the relationships among the inputs obvious

97 MATHEMATICS AND COMPUTING↗

Self-Organizing Maps and Their Applications to Data Analysis

Self-Organizing Maps(SOMs) are a form of unsupervised neural network that are used for visualization and exploratory data analysis of high dimensional datasets. Our goal was to understand how we can use a SOM to gain insights about datasets. We do this by first understanding the initialization, training, error metrics, and convergence properties of the SOM. Next we discuss the ways to interpret and visualize a Self-Organizing Map. Finally we used real datasets to understand what the Self-Organizing Map can tell us about labeled and unlabeled data. Based on experiments with our datasets we found that the Self-Organizing Map can tell us about the spacing and position of high dimensional clusters, help us find non-linear patterns, and give us insight into the shape of our data.

97 MATHEMATICS AND COMPUTING↗

Tensile properties, strain rate sensitivity, and activation volume of additively manufactured 316L stainless steels [Tensile properties of 316L stainless steels made by laser powder-bed-fusion additive manufacturing]

The tensile properties of additively manufactured (AM) metals and alloys are among the most important variables that impact the potential applications of these materials. Here we examine and report on the tensile properties of AM 316L stainless steels fabricated by the laser powder-bed-fusion (L-PBF) technique, via twelve sets of optimized laser processing parameters that produce materials with density >98.8 ± 0.10%. A heterogeneous microstructure is observed in all L-PBF samples, including microscopic features such as dislocations, cellular walls, elemental segregations, local misorientations, impurities, precipitates, and a large fraction of low-angle grain boundaries (2-10°, ~40–60%). The derived average grain size defined by high-angle grain boundaries (>10°) is ~30–50 μm. Tensile testing reveals a yield strength ranging from 552 to 635 MPa and a tensile-elongation-to-failure (TEF) of 0.09–0.42 for directly-printed samples, whereas these values are 592–690 MPa and 0.29–0.50 for samples machined from the as-built rectangular thin plates. In all samples, we observe a variation of tensile yield strength within ~15% but not the TEF, suggesting marginal microstructural changes despite a wide range of laser processing parameters. The large scatter of TEF in directly-printed samples originates from the sensitivity of thin gauge geometry (~2 mm 2 cross-section area) to the built-in flaws. We measured a substantially higher strain rate sensitivity (m~0.02–0.03) of L-PBF 316L compared to the coarse-grained counterparts (~0.006), together with a small activation volume of ~20–30b3 (where b is the Burgers vector of 316L). In conclusion, these deformation kinetics parameters suggest that the tensile plasticity of L-PBF 316L is controlled by a much finer microstructural length scale than the measured grain size, consistent with the high strength and juxtaposed nano- to macro-structures seen in these materials. Strategies to optimize the tensile properties of AM materials are discussed.

36 MATERIALS SCIENCE↗

Understanding the Effects of Tapering on Gaussian Process Regression

Gaussian process regression is a computationally expensive machine learning algorithm that requires the solution of a system of linear equations with a dense covariance matrix. We propose a modification that uses tapering to introduce sparsity into the covariance matrix followed by matrix reordering schemes to reduce the bandwidth of the sparse matrix. We also apply iterative refinement to our approximate solution in order to recover some of the accuracy lost to tapering. Our numerical tests showed that this modification works best with datasets that have many small values in the covariance matrix. This restricts the usefulness of the modification, so other techniques to reduce computational time may be worth considering in the future.

97 MATHEMATICS AND COMPUTING↗

Compressing unstructured mesh data from simulations using machine learning

The amount of data output from a computer simulation has grown to terabytes and petabytes as increasingly complex simulations are being run on massively parallel systems. As we approach exaflop computing in the next decade, it is expected that the I/O subsystem will not be able to write out these large volumes of data. In this paper, we explore the use of machine learning to compress the data before it is written out. Despite the computational constraints that limit us to using very simple learning algorithms, our results show that machine learning is a viable option for compressing unstructured data. Furthermore, we demonstrate that by simply using a better sampling algorithm to generate the training set, we can obtain more accurate results compared to random sampling, but at no extra cost. Further, by carefully selecting and incorporating points with high prediction error, we can improve reconstruction accuracy without sacrificing the compression rate.

97 MATHEMATICS AND COMPUTING↗

A processing diagram for high-density Ti-6Al-4V by selective laser melting

Density optimization is the first critical step in building additively manufactured parts with highquality and good mechanical properties. In earlier work, we developed an approach that combines simulations and experiments to identify processing parameters for high-density parts. Here we present the results for Ti-6Al-4V using the laser powder-bed-fusion technique. We show how simple, but approximate, simulations are used to design parameters for a limited set of single track experiments. The resulting melt-pool characteristics are then used to identify processing parameters for high-density pillars. From these experimental data, a processing diagram based on the normalized energy density concept is constructed, illustrating an optimized processing window for high- or low-density samples. Excellent mechanical properties are obtained for Ti-6Al-4V samples built from the optimized window. We discuss the limitations and implications of the processing diagram on the microstructure and mechanical properties.

36 MATERIALS SCIENCE↗

A Comparison of Compressed Sensing and Sparse Recovery Algorithms Applied to Simulation Data

The move toward exascale computing for scientific simulations is placing new demands on compression techniques. It is expected that the I/O system will not be able to support the volume of data that is expected to be written out. To enable quantitative analysis and scientific discovery, we are interested in techniques that compress high-dimensional simulation data and can provide perfect or near-perfect reconstruction. In this paper, we explore the use of compressed sensing (CS) techniques to reduce the size of the data before they are written out. Using large-scale simulation data, we investigate how the sufficient sparsity condition and the contrast in the data affect the quality of reconstruction and the degree of compression. Also, we provide suggestions for the practical implementation of CS techniques and compare them with other sparse recovery methods. Finally, our results show that despite longer times for reconstruction, compressed sensing techniques can provide near perfect reconstruction over a range of data with varying sparsity.

97 MATHEMATICS AND COMPUTING↗

Data mining and statistical inference in selective laser melting

Selective laser melting (SLM) is an additive manufacturing process that builds a complex three-dimensional part, layer-by-layer, using a laser beam to fuse fine metal powder together. The design freedom afforded by SLM comes associated with complexity. As the physical phenomena occur over a broad range of length and time scales, the computational cost of modeling the process is high. At the same time, the large number of parameters that control the quality of a part make experiments expensive. In this paper, we describe ways in which we can use data mining and statistical inference techniques to intelligently combine simulations and experiments to build parts with desired properties. We start with a brief summary of prior work in finding process parameters for high-density parts. We then expand on this work to show how we can improve the approach by using feature selection techniques to identify important variables, data-driven surrogate models to reduce computational costs, improved sampling techniques to cover the design space adequately, and uncertainty analysis for statistical inference. Here, our results indicate that techniques from data mining and statistics can complement those from physical modeling to provide greater insight into complex processes such as selective laser melting.

36 MATERIALS SCIENCE↗

Identifying and Exploiting Diurnal Motifs in Wind Generation Time Series Data

Wind energy is scheduled on the power grid using 0–6 h ahead forecasts generated from computer simulations or historical data. When the forecasts are inaccurate, control room operators use their expertise, as well as the actual generation from previous days, to estimate the amount of energy to schedule. However, this is a challenge, and it would be useful for the operators to have additional information they can exploit to make better informed decisions. In this paper, we use techniques from time series analysis to determine if there are motifs, or frequently occurring diurnal patterns in wind generation data. Here we compare two different representations of the data and four different ways of identifying the number of motifs. Using data from wind farms in Tehachapi Pass and mid-Columbia Basin, we describe our findings and discuss how these motifs can be used to guide scheduling decisions.

97 MATHEMATICS AND COMPUTING↗

Observation of keyhole-mode laser melting in laser powder-bed fusion additive manufacturing

Laser powder-bed fusion additive manufacturing of metals employs high-power focused laser beams. Typically, the depth of the molten pool is controlled by conduction of heat in the underlying solid material. But, under certain conditions, the mechanism of melting can change from conduction to so-called “keyhole-mode” laser melting. In this mode, the depth of the molten pool is controlled by evaporation of the metal. Keyhole-mode laser melting results in melt pool depths that can be much deeper than observed in conduction mode. In addition, the collapse of the vapor cavity that is formed by the evaporation of the metal can result in a trail of voids in the wake of the laser beam. In this paper, the experimental observation of keyhole-mode laser melting in a laser powder-bed fusion additive manufacturing setting for 316L stainless steel is presented. Finally, the conditions required to transition from conduction controlled melting to keyhole-mode melting are identified.

36 MATERIALS SCIENCE↗

Density of Additively-Manufactured, 316L SS Parts Using Laser Powder-Bed Fusion at Powers Up to 400W

Selective laser melting is a powder-based, additive-manufacturing process where a threedimensional part is produced, layer by layer, by using a high-energy laser beam to fuse the metallic powder particles. A particular challenge in this process is the selection of appropriate process parameters that result in parts with desired properties. In this study, we describe an approach to selecting parameters for high density (>99%) parts using 316L stainless steel. Though there has been significant success in achieving near-full density for 316L parts, this work has been limited to laser powers <225W. We discuss how we can exploit prior knowledge, design of computational experiments using a simple model of laser melting, and single-track experiments to determine the process parameters for use at laser powers up to 400W. Our results show that, at higher power values, there is a large range of scan speeds over which the relative density remains >99%, with the density reducing rapidly at high speeds due to insufficient melting, and less rapidly at low speeds due to the effect of voids created as the process enters keyhole mode.

316L stainless teel↗

Ensemble Feature Selection in Scientific Data Analysis

Feature selection has emerged as an important method of both examining and predicting results from scientific experiments. However, current feature selection methods cannot consistently provide meaningful descriptions of feature importance in the complex, high-dimensional, natural datasets encountered in scientific data analysis. Here, an ensemble feature selection method is demonstrated to collectively reduce the bias associated with individual feature selection algorithms and provide uncertainty estimates on feature importance ranks for both synthetic and natural datasets.

97 MATHEMATICS AND COMPUTING↗

A Block-Based MC-SURE Algorithm for Denoising Sensor Data Streams

We propose a strategy to automatically denoise sensor data streams corrupted with noise that can be approximated as additive white Gaussian noise. The proposed block-based method is adapted from the Monte-Carlo-SURE (MC-SURE) algorithm which enables the blind optimization of the denoising parameter of a wide class of filters. Our framework is formulated by identifying and addressing the challenges that arise when the MC-SURE algorithm is applied in an on-line data processing setting, where latency (and the length of data blocks) must be constrained. The strategy has been tested using real datasets. Our results indicate that the proposed method can be effectively used to handle the denoising of real sensor streaming data that reasonably fit the Gaussian model assumption.

97 MATHEMATICS AND COMPUTING↗