Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “performance tuning”

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 541 records · Page 30

Criticality analysis of nuclear binding energy neural networks

Machine learning methods, in particular deep learning methods such as artificial neural networks (ANNs) with many layers, have become widespread and useful tools in nuclear physics. However, these ANNs are typically treated as ‘black boxes’, with their architecture (width, depth, and weight/bias initialization) and the training algorithm and parameters chosen empirically by optimizing learning based on limited exploration. We test a non-empirical approach to understanding and optimizing nuclear physics ANNs by adapting a criticality analysis based on renormalization group flows in terms of the hyperparameters for weight/bias initialization, training rates, and the ratio of depth to width. This treatment utilizes the statistical properties of neural network initialization to find a generating functional for network outputs at any layer, allowing for a path integral formulation of the ANN outputs as a Euclidean statistical field theory. We use a prototypical example to test the applicability of this approach: a simple ANN for nuclear binding energies. We find that with training using a stochastic gradient descent optimizer, the predicted criticality behavior is realized, and optimal performance is found with critical tuning. However, the use of an adaptive learning algorithm leads to somewhat superior results without concern for tuning and thus obscures the analysis. Nevertheless, the criticality analysis offers a way to look within the black box of ANNs, which is a first step towards potential improvements in network performance beyond using adaptive optimizers.

artificial neural network↗

Advanced Modeling of Beam Physics and Performance Optimization for Nuclear Physics Colliders

High energy colliders provide a critical tool in nuclear physics study by probing the fundamental structure and dynamics of matter. To maximize the potential of scientific discovery in nuclear physics study, it is important to optimize the parameters of these colliders to attain the best performance. The performance of a collider is typically measured by its integrated luminosity of colliding beams since the probability of a new event is proportional to the integrated luminosity. However, the achievable luminosity is limited by the electromagnetic interactions (beam-beam effects) of two colliding beams at higher energy, and the interplay between the space-charge effects and the beam-beam effects at lower energy. To achieve the best performance of a collider means to attain the highest luminosity of the collider with optimized collider parameters. Optimizing the collider’s machine parameters is both computationally and experimentally expensive. A fast and robust computational framework including beam-beam and space-charge effects will be critical to attaining the best performance of the collider. In this project, we will study the beam dynamics challenges, specifically the interplay of the space-charge and the beam-beam effects, and the machine tuning models for maximizing the performance of RHIC experiments. We will develop an advanced modeling framework based on first-principles physical simulations, lattice models and the state-of-the-art machine learning methods and apply this framework to performance improvement of the RHIC in operation. We will build data manipulation packages to connect the simulation data and the experimental data with the framework, develop a self-consistent hybrid model of space-charge and beam-beam effects, study underlying physics mechanisms, build surrogate models using the labeled data, integrate the models into the advanced modeling framework, and apply the framework to RHIC luminosity (STAR and sPHENIX) optimization. The success of this project would substantially improve the performance of existing and future colliders and increase the opportunity for scientific discovery.

43 PARTICLE ACCELERATORS↗

A multiple objective optimization approach to quality control

The use of product quality as the performance criteria for manufacturing system control is explored. The goal in manufacturing, for economic reasons, is to optimize product quality. The problem is that since quality is a rather nebulous product characteristic, there is seldom an analytic function that can be used as a measure. Therefore standard control approaches, such as optimal control, cannot readily be applied. A second problem with optimizing product quality is that it is typically measured along many dimensions: there are many apsects of quality which must be optimized simultaneously. Very often these different aspects are incommensurate and competing. The concept of optimality must now include accepting tradeoffs among the different quality characteristics. These problems are addressed using multiple objective optimization. It is shown that the quality control problem can be defined as a multiple objective optimization problem. A controller structure is defined using this as the basis. Then, an algorithm is presented which can be used by an operator to interactively find the best operating point. Essentially, the algorithm uses process data to provide the operator with two pieces of information: (1) if it is possible to simultaneously improve all quality criteria, then determine what changes to the process input or controller parameters should be made to do this; and (2) if it is not possible to improve all criteria, and the current operating point is not a desirable one, select a criteria in which a tradeoff should be made, and make input changes to improve all other criteria. The process is not operating at an optimal point in any sense if no tradeoff has to be made to move to a new operating point. This algorithm ensures that operating points are optimal in some sense and provides the operator with information about tradeoffs when seeking the best operating point. The multiobjective algorithm was implemented in two different injection molding scenarios: tuning of process controllers to meet specified performance objectives and tuning of process inputs to meet specified quality objectives. Five case studies are presented.

Seaman, Christopher Michael↗

Experimental Platform for Ultra-high Dose Rate FLASH Irradiation of Small Animals Using a Clinical Linear Accelerator

Purpose: A key factor limiting the effectiveness of radiation therapy is normal tissue toxicity, and recent preclinical data have shown that ultra-high dose rate irradiation (>50 Gy/s, “FLASH”) potentially mitigates this effect. However, research in this field has been strongly limited by the availability of FLASH irradiators suitable for small animal experiments. We present a simple methodologic approach for FLASH electron small animal irradiation with a clinically available linear accelerator (LINAC). Methods and Materials: We investigated the FLASH irradiation potential of a Varian Clinac 21EX in both clinical mode and after tuning of the LINAC. We performed detailed FLUKA Monte Carlo and experimental dosimetric characterization at multiple experimental locations within the LINAC head. Results: Average dose rates of ≤74 Gy/s were achieved in clinical mode, and the dose rate after tuning exceeded 900 Gy/s. We obtained 220 Gy/s at 1-cm depth for a >4-cm field size with 90% homogeneity throughout a 2-cm-thick volume. Conclusions: We present an approach for using a clinical LINAC for FLASH irradiation. We obtained dose rates exceeding 200 Gy/s after simple tuning of the LINAC, with excellent dosimetric properties for small animal experiments. This will allow for increased availability of FLASH irradiation to the general research community.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

A Generalized Framework for Auto-tuning Stencil Computations

This work introduces a generalized framework for automatically tuning stencil computations to achieve superior performance on a broad range of multicore architectures. Stencil (nearest-neighbor) based kernels constitute the core of many important scientific applications involving block-structured grids. Auto-tuning systems search over optimization strategies to find the combination of tunable parameters that maximizes computational efficiency for a given algorithmic kernel. Although the auto-tuning strategy has been successfully applied to libraries, generalized stencil kernels are not amenable to packaging as libraries. Studied kernels in this work include both memory-bound kernels as well as a computation-bound bilateral filtering kernel. We introduce a generalized stencil auto-tuning framework that takes a straightforward Fortran expression of a stencil kernel and automatically generates tuned implementations of the kernel in C or Fortran to achieve performance portability across diverse computer architectures.

97 MATHEMATICS AND COMPUTING↗

Design of highly active and durable electrocatalysts by tuning nanostructure

Chemical-electrical energy conversion and storage are greatly attractive for the development of sustainable energy. Among the chemical-electrical energy conversion devices, polymer electrolyte membrane fuel cell (PEMFC) is one promising technology for portable applications, from as small as portable electronics, to as large as automobiles. One key for PEMFCs is the electrocatalyst. Development of high-performance electrocatalyst nanomaterials relies on tuning material structures at nanoscale. This is in particular manifested in the design of electrocatalysts demanding both high activity and durability. Through solution-phase-synthesis, high quality nanocrystals are prepared at various morphologies (shape-controlled) and compositions (composition-controlled). The nanostructure (i.e. selective exposing specific low-index surface, controlling core-shell structure, and etc.) of nanocrystals are also carefully tuned. Moreover, binary nanocrystal superlattices offer new opportunity to explore electrocatalysis by incorporating physical and chemical properties of different materials in one single, crystalline, and complex nanostructure. Using these highly controlled nanocrystals and nanostructures as research platforms, we explore the correlation between nanostructure and activity (and/or durability). Herein, a group of high performance electrocatalysts are developed for various electrocatalytic processes, including oxygen reduction reaction (ORR) and hydrogen evolution reaction (HER). It is well-demonstrated that the careful manipulation of nanostructure of electrocatalysts is crucial to improve the activity and durability.

Kang, Yijin↗

Magnetic Tuning of a Quasi-Periodic Elliptically Polarized Undulator

A 2.8 m long quasi-periodic APPLE II type (Advanced Planar Polarized Light Emitter II) undulator, is being commissioned at National Synchrotron Light Source II (NSLS-II). It will service the Electron SpectroMicroscopy (ESM) beamline in the framework of the NEXT (NSLS-II Experimental Tools) project [1]. It has been designated to provide high brilliance photon beams with variable polarization from VUV to soft X-Rays. Although this EPU (Elliptically Polarizing Undulator) was procured as a turn-key device from Kyma [2], the vendor was only responsible for the mechanical frame and the control system. Sorting and assembly of the magnet modules and the magnetic field tuning - Shimming and Magic Finger- were performed at the NSLS-II Magnetic Measurement Lab. In this report the field optimization techniques implemented in-house to correct the undesirable phasedependent errors and the final results of the magnetic tuning will be described together with the spectral performance of the device.

36 MATERIALS SCIENCE↗

Effective Carrier‐Concentration Tuning of SnO 2 Quantum Dot Electron‐Selective Layers for High‐Performance Planar Perovskite Solar Cells

Abstract The carrier concentration of the electron‐selective layer (ESL) and hole‐selective layer can significantly affect the performance of organic–inorganic lead halide perovskite solar cells (PSCs). Herein, a facile yet effective two‐step method, i.e., room‐temperature colloidal synthesis and low‐temperature removal of additive (thiourea), to control the carrier concentration of SnO 2 quantum dot (QD) ESLs to achieve high‐performance PSCs is developed. By optimizing the electron density of SnO 2 QD ESLs, a champion stabilized power output of 20.32% for the planar PSCs using triple cation perovskite absorber and 19.73% for those using CH 3 NH 3 PbI 3 absorber is achieved. The superior uniformity of low‐temperature processed SnO 2 QD ESLs also enables the fabrication of ≈19% efficiency PSCs with an aperture area of 1.0 cm 2 and 16.97% efficiency flexible device. The results demonstrate the promise of carrier‐concentration‐controlled SnO 2 QD ESLs for fabricating stable, efficient, reproducible, large‐scale, and flexible planar PSCs.

Yang, Guang↗

Tuning the electronic structure of Ag-Pd alloys to enhance performance for alkaline oxygen reduction

Alloying is a powerful tool that can improve the electrocatalytic performance and viability of diverse electrochemical renewable energy technologies. Herein, we enhance the activity of Pd-based electrocatalysts via Ag-Pd alloying while simultaneously lowering precious metal content in a broad-range compositional study focusing on highly comparable Ag-Pd thin films synthesized systematically via electron-beam physical vapor co-deposition. Cyclic voltammetry in 0.1 M KOH shows enhancements across a wide range of alloys; even slight alloying with Ag (e.g. Ag 0.1 Pd 0.9 ) leads to intrinsic activity enhancements up to 5-fold at 0.9 V vs. RHE compared to pure Pd. Based on density functional theory and x-ray absorption, we hypothesize that these enhancements arise mainly from ligand effects that optimize adsorbate–metal binding energies with enhanced Ag-Pd hybridization. This work shows the versatility of coupled experimental-theoretical methods in designing materials with specific and tunable properties and aids the development of highly active electrocatalysts with decreased precious-metal content.

25 ENERGY STORAGE↗

Tuning Superhydrophilic Nanostructured Surfaces to Maximize Water Droplet Evaporation Heat Transfer Performance

Spraying water droplets on air fin surfaces is often used to augment the performance of air-cooled Rankine power plant condensers and wet cooling tower heat exchangers for building air-conditioning systems. To get the best performance in such processes, the water droplets delivered to the surface should spread rapidly into an extensive, thin film and evaporate with no liquid leaving the surface due to recoil or splashing. This paper presents predictions of theoretical/computational modeling and results of experimental studies of droplet spreading on thin-layer, nanostructured, superhydrophilic surfaces that exhibit very high wicking rates (wickability) in the porous layer. Analysis of the experimental data in the model framework illuminates the key aspects of the physics of the droplet-spreading process and evaporation heat transfer. This analysis also predicts the dependence of droplet-spreading characteristics on the nanoporous surface morphology and other system parameters. The combined results of this investigation indicate specific key strategies for design and fabrication of surface coatings that will maximize the heat transfer performance for droplet evaporation on heat exchanger surfaces. The implications regarding wickability effects on pool boiling processes are also discussed.

Engineering↗

Optimization of a Lattice Boltzmann Computation on State-of-the-Art Multicore Platforms

We present an auto-tuning approach to optimize application performance on emerging multicore architectures. The methodology extends the idea of search-based performance optimizations, popular in linear algebra and FFT libraries, to application-specific computational kernels. Our work applies this strategy to a lattice Boltzmann application (LBMHD) that historically has made poor use of scalar microprocessors due to its complex data structures and memory access patterns. We explore one of the broadest sets of multicore architectures in the HPC literature, including the Intel Xeon E5345 (Clovertown), AMD Opteron 2214 (Santa Rosa), AMD Opteron 2356 (Barcelona), Sun T5140 T2+ (Victoria Falls), as well as a QS20 IBM Cell Blade. Rather than hand-tuning LBMHD for each system, we develop a code generator that allows us to identify a highly optimized version for each platform, while amortizing the human programming effort. Results show that our auto-tuned LBMHD application achieves up to a 15x improvement compared with the original code at a given concurrency. Additionally, we present detailed analysis of each optimization, which reveal surprising hardware bottlenecks and software challenges for future multicore systems and applications.

97 MATHEMATICS AND COMPUTING↗

Graphene nanopore field effect transistors

Graphene holds great promise for replacing conventional Si material in field effect transistors (FETs) due to its high carrier mobility. Previously proposed graphene FETs either suffer from low ON-state current resulting from constrained channel width or require complex fabrication processes for edge-defecting or doping. Here, we propose an alternative graphene FET structure created on intrinsic metallic armchair-edged graphene nanoribbons with uniform width, where the channel region is made semiconducting by drilling a pore in the interior, and the two ends of the nanoribbon act naturally as connecting electrodes. The proposed GNP-FETs have high ON-state currents due to seamless atomic interface between the channel and electrodes and are able to be created with arbitrarily wide ribbons. In addition, the performance of GNP-FETs can be tuned by varying pore size and ribbon width. As a result, their performance and fabrication process are more predictable and controllable in comparison to schemes based on edge-defects and doping. Using first-principle transport calculations, we show that GNP-FETs can achieve competitive leakage current of ∼70 pA, subthreshold swing of ∼60 mV/decade, and significantly improved On/Off current ratios on the order of 10{sup 5} as compared with other forms of graphene FETs.

77 NANOSCIENCE AND NANOTECHNOLOGY↗

Evaluation of NorESM-OC (versions 1 and 1.2), the ocean carbon-cycle stand-alone configuration of the Norwegian Earth System Model (NorESM1)

Idealised and hindcast simulations performed with the stand-alone ocean carbon-cycle configuration of the Norwegian Earth System Model (NorESM-OC) are described and evaluated. We present simulation results of three different model configurations (two different model versions at different grid resolutions) using two different atmospheric forcing data sets. Model version NorESM-OC1 corresponds to the version that is included in the NorESM-ME1 fully coupled model, which participated in CMIP5. The main update between NorESM-OC1 and NorESM-OC1.2 is the addition of two new options for the treatment of sinking particles. We find that using a constant sinking speed, which has been the standard in NorESM's ocean carbon cycle module HAMOCC (HAMburg Ocean Carbon Cycle model), does not transport enough particulate organic carbon (POC) into the deep ocean below approximately 2000 m depth. The two newly implemented parameterisations, a particle aggregation scheme with prognostic sinking speed, and a simpler scheme that uses a linear increase in the sinking speed with depth, provide better agreement with observed POC fluxes. Additionally, reduced deep ocean biases of oxygen and remineralised phosphate indicate a better performance of the new parameterisations. For model version 1.2, a re-tuning of the ecosystem parameterisation has been performed, which (i) reduces previously too high primary production at high latitudes, (ii) consequently improves model results for surface nutrients, and (iii) reduces alkalinity and dissolved inorganic carbon biases at low latitudes. We use hindcast simulations with prescribed observed and constant (pre-industrial) atmospheric CO 2 concentrations to derive the past and contemporary ocean carbon sink. As a result, for the period 1990–1999 we find an average ocean carbon uptake ranging from 2.01 to 2.58 Pg C yr -1 depending on model version, grid resolution, and atmospheric forcing data set.

54 ENVIRONMENTAL SCIENCES↗

FPGA-based HPC accelerators: An evaluation on performance and energy efficiency

Hardware specialization is a promising direction for the future of digital computing. Reconfigurable technologies enable hardware specialization with modest non-recurring engineering cost, but their performance and energy efficiency compared to state-of-the-art processor architectures remain an open question. In this article, we use FPGAs to evaluate the benefits of building specialized hardware for numerical kernels found in scientific applications. In order to properly evaluate performance, we not only compare Intel Arria 10 and Xilinx U280 performance against Intel Xeon, Intel Xeon Phi, and NVIDIA V100 GPUs, but we also extend the Empirical Roofline Toolkit (ERT) to FPGAs in order to assess our results in terms of the Roofline model. We show design optimization and tuning techniques for peak FPGA performance at reasonable hardware usage and power consumption. As FPGA peak performance is known to be far less than that of a GPU, we also benchmark the energy efficiency of each platform for the scientific kernels comparing against microbenchmark and technological limits. Results show that while FPGAs struggle to compete in absolute terms with GPUs on memory- and compute-intensive kernels, they require far less power and can deliver nearly the same energy efficiency.

97 MATHEMATICS AND COMPUTING↗

Tuning the Surface Chemistry of Metal Organic Framework Nodes: Proton Topology of the Metal-Oxide-Like Zr 6 Nodes of UiO-66 and NU-1000

Some metal organic frameworks (MOFs) incorporate nodes that are nanoscale metal oxides, and the hydroxycontaining functional groups on them provide opportunities for introducing catalytic sites with precisely defined structures. Investigations have been done to understand the structures of these groups on nodes and node vacancies, because, in prospect, atomic-scale modulation of the composition, areal density, and/or siting of the groups would open up possibilities for exquisite tuning of the siting and performance of subsequently anchored catalytic units (e.g., single metal ions, pairs of metal ions, or well-defined metal-ion-containing clusters). In this work we have combined infrared (IR) spectroscopy and density functional theory (DFT) to demonstrate tuning of these sites, namely, hydrogen-bonded OH/OH 2 groups on the Zr 6 nodes of the MOFs UiO-66 and NU-1000 via the intermediacy of node methoxy (or ethoxy) groups formed from methanol (or ethanol). Methoxy (or ethoxy) groups on node vacancy sites are converted to a structure incorporating one vacant Zr site and one terminal OH group per face by reaction with water. Our results highlight how the combination of DFT and IR spectroscopy facilitates the determination of the identity and chemistry of the functional groups on MOF node vacancies and defect sites.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

CW laser strategies for multi-parameter measurements of high-speed flows containing either NO or O2

Measurements of gasdynamic quantities were performed using a rapid-tuning CW dye laser to resolve Doppler-shifted spectral features in either the O2 Schumann-Runge bands or the NO gamma band near 225 nm. With the rapid-tuning capability, spectral features were acquired at a repetition rate of 4 kHz. Monitoring O2 transitions provided estimates of velocities while monitoring collision-broadened NO line pairs provided simultaneous measurements of velocity, temperature and pressure. Experiments were first performed in absorption within the transient one-dimensional flows generated in a shock tube. Agreement between measured and theoretical values, as calculated from one-dimensional shock relations, was typically better than 5 percent. The method was extended to fluorescence detection of NO in a static cell. Temperature and pressure were extracted from recorded profiles, and the results agreed well with expected values.

Dirosa, M. D.↗