Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Dynamic Line Rating”

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 433 records · Page 24

OBSERVATIONS OF STARBURST GALAXIES WITH FAR-ULTRAVIOLET SPECTROGRAPHIC EXPLORER: GALACTIC FEEDBACK IN THE LOCAL UNIVERSE

We have analyzed FUSE (905-1187 A) spectra of a sample of 16 local starburst galaxies. These galaxies cover almost three orders of magnitude in star-formation rates and over two orders of magnitude in stellar mass. Absorption features from the stars and interstellar medium are observed in all the spectra. The strongest interstellar absorption features are generally blue-shifted by {approx} 50-300 km s{sup -1}, implying the almost ubiquitous presence of starburst-driven galactic winds in this sample. The outflow velocites increase with both the star-formation rate and the star-formation rate per unit stellar mass, consistent with a galactic wind, driven by the population of massive stars. We find outflowing coronal-phase gas (T {approx}10{sup 5.5} K) detected via the O VI absorption line in nearly every galaxy. The O VI absorption-line profile is optically thin, is generally weak near the galaxy-systemic velocity, and has a higher mean outflow velocity than seen in the lower ionization lines. The relationship between the line width and column density for the O VI absorbing gas is in good agreement with expectations for radiatively cooling and outflowing gas. Such gas will be created in the interaction of the hot out-rushing wind seen in X-ray emission and cool dense ambient material. O VI emission is not generally detected in our sample, suggesting that radiative cooling by the coronal gas is not dynamically significant in draining energy from galactic winds. We find that the measured outflow velocities in the H I and H II phases of the interstellar gas in a given galaxy increase with the strength (equivalent width) of the absorption feature and not with the ionization potential of the species. The strong lines often have profiles consisting of a broad and optically-thick component centered near the galaxy-systemic velocity and weaker but highly blue-shifted absorption. This suggests that the outflowing gas with high velocity has a lower column density than the more quiescent gas, and can only be readily detected in the strongest absorption lines. From direct observations below the Lyman edge in the galaxy rest frame, we find no evidence of Lyman continuum radiation escaping from any of the galaxies in the sample. Moreover, the small escape fraction of light in the center of the strong C II absorption feature confirms the high opacity below the Lyman limit in the neutral ISM. The absolute fraction of escaping Lyman continuum photons is typically <1%. This sample provides a unique window on the global properties of local star-forming galaxies as observed in the far-UV and also provides a useful comparison sample for understanding spectra of high-redshift star-forming galaxies.

79 ASTRONOMY AND ASTROPHYSICS↗

STRUCTURE AND DYNAMICS OF QUIESCENT FILAMENT CHANNELS OBSERVED BY HINODE/XRT AND STEREO/EUVI

We present a study of the structure and dynamics of quiescent filament channels observed by Hinode/XRT and STEREO/EUVI at the solar minimum 23/24 from 2006 November to 2008 December. For 12 channels identified on the solar disk (Group I channels), we find that the morphology of the structure on the two sides of the channel is asymmetric in both X-rays and EUV: the eastern side has curved features while the western side has straight features. We interpret the results in terms of a magnetic flux rope model. The asymmetry in the morphology is due to the variation in axial flux of the flux rope along the channel, which causes the field lines from one polarity to turn into the flux rope (curved feature), while the field lines from the other polarity are connected to very distant sources (straight). For most of the 68 channels identified by cavities at the east and west limbs (Group II channels), the asymmetry cannot be clearly identified, which is likely due to the fact that the axial flux may be relatively constant along such channels. Corresponding cavities are identified only for 5 of the 12 Group I channels, while Group II channels are identified for all of the 68 cavity pairs. The studied filament channels are often observed as dark channels in X-rays and EUV. Sheared loops within Group I channels are often seen in X-rays, but are rarely seen in Group II channels as shown in the X-ray Telescope daily synoptic observations. A survey of the dynamics of studied filament channels shows that filament eruptions occur at an average rate of 1.4 filament eruptions per channel per solar rotation.

79 ASTRONOMY AND ASTROPHYSICS↗

Heating and dynamics of two flare loop systems observed by AIA and EIS

We investigate heating and evolution of flare loops in a C4.7 two-ribbon flare on 2011 February 13. From Solar Dynamics Observatory/Atmospheric Imaging Assembly (AIA) imaging observations, we can identify two sets of loops. Hinode/EUV Imaging Spectrometer (EIS) spectroscopic observations reveal blueshifts at the feet of both sets of loops. The evolution and dynamics of the two sets are quite different. The first set of loops exhibits blueshifts for about 25 minutes followed by redshifts, while the second set shows stronger blueshifts, which are maintained for about one hour. The UV 1600 observation by AIA also shows that the feet of the second set of loops brighten twice. These suggest that continuous heating may be present in the second set of loops. We use spatially resolved UV light curves to infer heating rates in the few tens of individual loops comprising the two loop systems. With these heating rates, we then compute plasma evolution in these loops with the 'enthalpy-based thermal evolution of loops' model. The results show that, for the first set of loops, the synthetic EUV light curves from the model compare favorably with the observed light curves in six AIA channels and eight EIS spectral lines, and the computed mean enthalpy flow velocities also agree with the Doppler shift measurements by EIS. For the second set of loops modeled with twice-heating, there are some discrepancies between modeled and observed EUV light curves in low-temperature bands, and the model does not fully produce the prolonged blueshift signatures as observed. We discuss possible causes for the discrepancies.

79 ASTRONOMY AND ASTROPHYSICS↗

Kepler-413B: A slightly misaligned, Neptune-size transiting circumbinary planet

We report the discovery of a transiting, R{sub p} = 4.347 ± 0.099R {sub ⊕}, circumbinary planet (CBP) orbiting the Kepler K+M eclipsing binary (EB) system KIC 12351927 (Kepler-413) every ∼66 days on an eccentric orbit with a{sub p} = 0.355 ± 0.002 AU, e{sub p} = 0.118 ± 0.002. The two stars, with M{sub A} = 0.820 ± 0.015 M {sub ☉}, R{sub A} = 0.776 ± 0.009 R {sub ☉} and M{sub B} = 0.542 ± 0.008 M {sub ☉}, R{sub B} = 0.484 ± 0.024 R {sub ☉}, respectively, revolve around each other every 10.11615 ± 0.00001 days on a nearly circular (e {sub EB} = 0.037 ± 0.002) orbit. The orbital plane of the EB is slightly inclined to the line of sight (i {sub EB} = 87.°33 ± 0.°06), while that of the planet is inclined by ∼2.°5 to the binary plane at the reference epoch. Orbital precession with a period of ∼11 yr causes the inclination of the latter to the sky plane to continuously change. As a result, the planet often fails to transit the primary star at inferior conjunction, causing stretches of hundreds of days with no transits (corresponding to multiple planetary orbital periods). We predict that the next transit will not occur until 2020. The orbital configuration of the system places the planet slightly closer to its host stars than the inner edge of the extended habitable zone. Additionally, the orbital configuration of the system is such that the CBP may experience Cassini State dynamics under the influence of the EB, in which the planet's obliquity precesses with a rate comparable to its orbital precession. Depending on the angular precession frequency of the CBP, it could potentially undergo obliquity fluctuations of dozens of degrees (and complex seasonal cycles) on precession timescales.

79 ASTRONOMY AND ASTROPHYSICS↗

MOSFIRE ABSORPTION LINE SPECTROSCOPY OF z > 2 QUIESCENT GALAXIES: PROBING A PERIOD OF RAPID SIZE GROWTH

Using the MOSFIRE near-infrared multi-slit spectrograph on the Keck 1 Telescope, we have secured high signal-to-noise ratio absorption line spectra for six massive galaxies with redshift 2 < z < 2.5. Five of these galaxies lie on the red sequence and show signatures of passive stellar populations in their rest-frame optical spectra. By fitting broadened spectral templates we have determined stellar velocity dispersions and, with broad-band Hubble Space Telescope and Spitzer photometry and imaging, stellar masses and effective radii. Using this enlarged sample of galaxies, we confirm earlier suggestions that quiescent galaxies at z > 2 have small sizes and large velocity dispersions compared to local galaxies of similar stellar mass. The dynamical masses are in very good agreement with stellar masses (log M {sub *}/M {sub dyn} = –0.02 ± 0.03), although the average stellar-to-dynamical mass ratio is larger than that found at lower redshift (–0.23 ± 0.05). By assuming evolution at fixed velocity dispersion, not only do we confirm a surprisingly rapid rate of size growth but we also consider the necessary evolutionary track on the mass-size plane and find a slope α = dlog R{sub e} /dlog M {sub *} ≳ 2 inconsistent with most numerical simulations of minor mergers. Both results suggest an additional mechanism may be required to explain the size growth of early galaxies.

79 ASTRONOMY AND ASTROPHYSICS↗

A STRONGLY COUPLED REACTOR CORE ISOLATION COOLING SYSTEM MODEL FOR EXTENDED STATION BLACK-OUT ANALYSES

The reactor core isolation cooling (RCIC) system in a boiling water reactor (BWR) provides makeup cooling water to the reactor pressure vessel (RPV) when the main steam lines are isolated and the normal supply of water to the reactor vessel is lost. The RCIC system operates independently of AC power, service air, or external cooling water systems. The only required external energy source is from the battery to maintain the logic circuits to control the opening and/or closure of valves in the RCIC systems in order to control the RPV water level by shutting down the RCIC pump to avoid overfilling the RPV and flooding the steam line to the RCIC turbine. It is generally considered in almost all the existing station black-out accidents (SBO) analyses that loss of the DC power would result in overfilling the steam line and allowing liquid water to flow into the RCIC turbine, where it is assumed that the turbine would then be disabled. This behavior, however, was not observed in the Fukushima Daiichi accidents, where the Unit 2 RCIC functioned without DC power for nearly three days. Therefore, more detailed mechanistic models for RCIC system components are needed to understand the extended SBO for BWRs. As part of the effort to develop the next generation reactor system safety analysis code RELAP-7, we have developed a strongly coupled RCIC system model, which consists of a turbine model, a pump model, a check valve model, a wet well model, and their coupling models. Unlike the traditional SBO simulations where mass flow rates are typically given in the input file through time dependent functions, the real mass flow rates through the turbine and the pump loops in our model are dynamically calculated according to conservation laws and turbine/pump operation curves. A simplified SBO demonstration RELAP-7 model with this RCIC model has been successfully developed. The demonstration model includes the major components for the primary system of a BWR, as well as the safety system components such as the safety relief valve (SRV), the RCIC system, the wet well, and the dry well. The results show reasonable system behaviors while exhibiting rich dynamics such as variable flow rates through RCIC turbine and pump during the SBO transient. The model has the potential to resolve the Fukushima RCIC mystery after adding the off-design two-phase turbine operation model and other additional improvements.

RCIC↗

MULTI-WAVELENGTH OBSERVATIONS OF COMET C/2011 L4 (PAN-STARRS)

The dynamically new comet C/2011 L4 (Pan-STARRS) is one of the brightest comets observed since the great comet C/1995 O1 (Hale-Bopp). Here, we present our multi-wavelength observations of C/2011 L4 during its in-bound passage to the inner solar system. A strong absorption band of water ice at 2.0 μm was detected in the near-infrared spectra, obtained with the 8 m Gemini-North and 3 m Infrared Telescope Facility Telescopes. The companion 1.5 μm band of water ice, however, was not observed. Spectral modeling shows that the absence of the 1.5 μm feature can be explained by the presence of sub-micron-sized fine ice grains. No gas lines (i.e., CN, HCN, or CO) were observed pre-perihelion in either the optical or the submillimeter. We derived 3σ upper limits for the CN and CO production rates. The comet exhibited a very strong continuum in the optical and its slope seemed to become redder as the comet approached the Sun. Our observations suggest that C/2011 L4 is an unusually dust-rich comet with a dust-to-gas mass ratio >4.

79 ASTRONOMY AND ASTROPHYSICS↗

STAR FORMATION IN TWO LUMINOUS SPIRAL GALAXIES

We examined star formation in two very luminous (M{sub V} = –22 to –23) Sc-type spiral galaxies, NGC 801 and UGC 2885, using ultra-deep Hα images. We combine these Hα images with UBV and Two-Micron All-Sky Survey JHK images and H I maps to explore the star formation characteristics of disk galaxies at high luminosity. Hα traces star formation in these galaxies to 4-6 disk scale lengths, but the lack of detection of Hα further out is likely due to the loss of Lyman continuum photons. Considering gravitational instabilities alone, we find that the gas and stars in the outer regions are marginally stable in an average sense, but considering dissipative gas and radial and azimuthal forcing, the outer regions are marginally unstable to forming spiral arms. Star formation is taking place in spiral arms, which are regions of locally higher gas densities. Furthermore, we have traced smooth exponential stellar disks over four magnitudes in V-band surface brightness and 4-6 disk scale lengths, in spite of a highly variable gravitational instability parameter. Thus, gravitational instability thresholds do not seem relevant to the stellar disk. One possibility for creating an exponential disk is that the molecular cloud densities and star formation rates have exponential profiles and this fact forces the stellar disk to build up such a profile. Another possibility is that the stellar disk is continuously adjusted to an exponential shape regardless of the star formation profile, for example, through global dynamical processes that scatter stars. However, such scattering processes are only known to operate in spiral systems, in which case they cannot explain the same dilemma of smooth exponential disks observed in dwarf irregular galaxies.

79 ASTRONOMY AND ASTROPHYSICS↗

THE EVOLUTION OF CLOUD CORES AND THE FORMATION OF STARS

For a number of starless cores, self-absorbed molecular line and column density observations have implied the presence of large-amplitude oscillations. We examine the consequences of these oscillations on the evolution of the cores and the interpretation of their observations. We find that the pulsation energy helps support the cores and that the dissipation of this energy can lead toward instability and star formation. In this picture, the core lifetimes are limited by the pulsation-decay timescales, dominated by non-linear mode-mode coupling, and on the order of {approx_equal} few x 10{sup 5}-10{sup 6} yr. Notably, this is similar to what is required to explain the relatively low rate of conversion of cores into stars. For cores with large-amplitude oscillations, dust continuum observations may appear asymmetric or irregular. As a consequence, some of the cores that would be classified as super-critical may be dynamically stable when oscillations are taken into account. Thus, our investigation motivates a simple hydrodynamic picture, capable of reproducing many of the features of the progenitors of stars without the inclusion of additional physical processes, such as large-scale magnetic fields.

79 ASTRONOMY AND ASTROPHYSICS↗

THERMAL AND DYNAMICAL PROPERTIES OF GAS ACCRETING ONTO A SUPERMASSIVE BLACK HOLE IN AN ACTIVE GALACTIC NUCLEUS

We study stability of gas accretion in active galactic nuclei (AGNs). Our grid-based simulations cover a radial range from 0.1 to 200 pc, which may enable linking the galactic/cosmological simulations with small-scale black hole (BH) accretion models within a few hundreds of Schwarzschild radii. Here, as in previous studies by our group, we include gas radiative cooling as well as heating by a sub-Eddington X-ray source near the central supermassive BH of 10{sup 8} M{sub Sun }. Our theoretical estimates and simulations show that for the X-ray luminosity, L{sub X} {approx} 0.008 L{sub Edd}, the gas is thermally and convectively unstable within the computational domain. In the simulations, we observe that very tiny fluctuations in an initially smooth, spherically symmetric, accretion flow, grow first linearly and then nonlinearly. Consequently, an initially one-phase flow relatively quickly transitions into a two-phase/cold-hot accretion flow. For L{sub X} = 0.015 L{sub Edd} or higher, the cold clouds continue to accrete but in some regions of the hot phase, the gas starts to move outward. For L{sub X} < 0.015 L{sub Edd}, the cold phase contribution to the total mass accretion rate only moderately dominates over the hot phase contribution. This result might have some consequences for cosmological simulations of the so-called AGN feedback problem. Our simulations confirm the previous results of Barai et al. who used smoothed particle hydrodynamic (SPH) simulations to tackle the same problem. Here, however, because we use a grid-based code to solve equations in one dimension and two dimensions, we are able to follow the gas dynamics at much higher spacial resolution and for longer time compared with the three-dimensional SPH simulations. One of the new features revealed by our simulations is that the cold condensations in the accretion flow initially form long filaments, but at the later times, those filaments may break into smaller clouds advected outward within the hot outflow. Therefore, these simulations may serve as an attractive model for the so-called narrow-line region in AGNs.

79 ASTRONOMY AND ASTROPHYSICS↗

Simulations of classical three-body thermalization in one dimension

One-dimensional systems, such as nanowires or electrons moving along strong magnetic field lines, have peculiar thermalization physics. The binary collision of pointlike particles, typically the dominant process for reaching thermal equilibrium in higher-dimensional systems, cannot thermalize a 1D system. We study how dilute classical 1D gases thermalize through three-body collisions. We consider a system of identical classical point particles with pairwise repulsive inverse power-law potential V ij ∝ 1/|x i –x j | n or the pairwise Lennard-Jones potential. Using Monte Carlo methods, we compute a collision kernel and use it in the Boltzmann equation to evolve a perturbed thermal state with temperature T toward equilibrium. We explain the shape of the kernel and its dependence on the system parameters. Additionally, we implement molecular dynamics simulations of a many-body gas and show agreement with the Boltzmann evolution in the low-density limit. For the inverse power-law potential, the rate of thermalization is proportional to ρ 2 ⁢T$\frac{1}{2}$ – $\frac{1}{n}$, where ρ is the number density. Furthermore, the corresponding proportionality constant decreases with increasing n.

1-dimensional systems↗

Field Test of Boiler Primary Loop Temperature Controller

Beyond these initial system efficiency upgrades are an emerging class of Advanced Load Monitoring (ALM) aftermarket controllers that dynamically respond to the boiler load, with claims of 10% to 30% of fuel savings over a heating season. For hydronic boilers specifically, these devices perform load monitoring, with continuous measurement of supply and in some cases return water temperatures. Energy savings from these ALM controllers are derived from dynamic management of the boiler differential, where a microprocessor with memory of past boiler cycles prevents the boiler from firing for a period of time, to limit cycling losses and inefficient operation during perceived low load conditions. These differ from OTR controllers, which vary boiler setpoint temperatures with ambient conditions while maintaining a fixed differential. PARR installed and monitored the performance of one type of ALM controller, the M2G from Greffen Systems, at multifamily sites in the city of Chicago and its suburb Cary, IL, both with existing OTR control. Results show that energy savings depend on the degree to which boilers are over-sized for their load, represented by cycling rates. Also savings vary over the heating season with cycling rates, with greater savings observed in shoulder months. Over the monitoring period, over-sized boilers at one site showed reductions in cycling and energy consumption in line with prior laboratory studies, while less over-sized boilers at another site showed muted savings.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Building America Case Study: Advanced Boiler Load Monitoring Controllers, Chicago, Illinois (Fact Sheet)

Beyond these initial system efficiency upgrades are an emerging class of Advanced Load Monitoring (ALM) aftermarket controllers that dynamically respond to the boiler load, with claims of 10% to 30% of fuel savings over a heating season. For hydronic boilers specifically, these devices perform load monitoring, with continuous measurement of supply and in some cases return water temperatures. Energy savings from these ALM controllers are derived from dynamic management of the boiler differential, where a microprocessor with memory of past boiler cycles prevents the boiler from firing for a period of time, to limit cycling losses and inefficient operation during perceived low load conditions. These differ from OTR controllers, which vary boiler setpoint temperatures with ambient conditions while maintaining a fixed differential. PARR installed and monitored the performance of one type of ALM controller, the M2G from Greffen Systems, at multifamily sites in the city of Chicago and its suburb Cary, IL, both with existing OTR control. Results show that energy savings depend on the degree to which boilers are over-sized for their load, represented by cycling rates. Also savings vary over the heating season with cycling rates, with greater savings observed in shoulder months. Over the monitoring period, over-sized boilers at one site showed reductions in cycling and energy consumption in line with prior laboratory studies, while less over-sized boilers at another site showed muted savings.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

NON-EQUILIBRIUM CHEMISTRY OF DYNAMICALLY EVOLVING PRESTELLAR CORES. II. IONIZATION AND MAGNETIC FIELD

We study the effect that non-equilibrium chemistry in dynamical models of collapsing molecular cloud cores has on measurements of the magnetic field in these cores, the degree of ionization, and the mean molecular weight of ions. We find that OH and CN, usually used in Zeeman observations of the line-of-sight magnetic field, have an abundance that decreases toward the center of the core much faster than the density increases. As a result, Zeeman observations tend to sample the outer layers of the core and consistently underestimate the core magnetic field. The degree of ionization follows a complicated dependence on the number density at central densities up to 10{sup 5} cm{sup -3} for magnetic models and 10{sup 6} cm{sup -3} in non-magnetic models. At higher central densities, the scaling approaches a power law with a slope of -0.6 and a normalization which depends on the cosmic-ray ionization rate {zeta} and the temperature T as ({zeta}T){sup 1/2}. The mean molecular weight of ions is systematically lower than the usually assumed value of 20-30, and, at high densities, approaches a value of 3 due to the asymptotic dominance of the H{sup +}{sub 3} ion. This significantly lower value implies that ambipolar diffusion operates faster.

79 ASTRONOMY AND ASTROPHYSICS↗

VARIABILITY OF DISK EMISSION IN PRE-MAIN SEQUENCE AND RELATED STARS. III. EXPLORING STRUCTURAL CHANGES IN THE PRE-TRANSITIONAL DISK IN HD 169142

We present near-IR (NIR) and far-UV observations of the pre-transitional (gapped) disk in HD 169142 using NASA's Infrared Telescope Facility and Hubble Space Telescope. The combination of our data along with existing data sets into the broadband spectral energy distribution reveals variability of up to 45% between ∼1.5-10 μm over a maximum timescale of 10 yr. All observations known to us separate into two distinct states corresponding to a high near-IR state in the pre-2000 epoch and a low state in the post-2000 epoch, indicating activity within the ≲1 AU region of the disk. Through analysis of the Pa β and Br γ lines in our data we derive a mass accretion rate in 2013 May of M-dot ≈ (1.5-2.7) × 10{sup –9} M {sub ☉} yr{sup –1}. We present a theoretical modeling analysis of the disk in HD 169142 using Monte-Carlo radiative transfer simulation software to explore the conditions and perhaps signs of planetary formation in our collection of 24 yr of observations. We find that shifting the outer edge (r ≈ 0.3 AU) of the inner disk by 0.05 AU toward the star (in simulation of accretion and/or sculpting by forming planets) successfully reproduces the shift in NIR flux. We establish that the ∼40-70 AU dark ring imaged in the NIR by Quanz et al. and Momose et al. and at 7 mm by Osorio et al. may be reproduced with a 30% scaled density profile throughout the region, strengthening the link to this structure being dynamically cleared by one or more planetary mass bodies.

79 ASTRONOMY AND ASTROPHYSICS↗

A molecular dynamics study of collisional heat transfer to nanoclusters in the gas phase

Nanoclusters in the gas phase grow by condensation and coagulation, which heat up nanoclusters, creating thermal non-equilibrium with the surrounding gas. The extent of non-equilibrium for nanoclusters is dependent upon the rate of heating relative to the rate of cooling brought about by nanocluster-gas molecule collisions, hence the latter process plays an important role in nanocluster energy evolution during growth. We apply molecular dynamics simulations to investigate heat transfer between metal nanoclusters and gas molecules relevant to high temperature aerosol systems. In analysis, we first define a thermal reemission coefficient (TRC), which is a correction factor for the portion of energy possessed by reemitted gas molecules after collision. Here, the TRC differs from the thermal accommodation coefficient (TAC), which is a correction factor for the maximum possible energy transfer rate from nanoclusters to gas molecules based on the diffusive collision model. We find the TRC is insensitive to nanocluster size but is affected by the nanocluster to gas temperature ratio and the nanocluster to gas atomic mass ratio. Comparison to the traditionally-defined TAC shows that the TAC is insensitive to temperature, a phenomenon that has been widely indicated in literature but not quantitatively explained. We show how the temperature insensitivity of the TAC arises because of the manner in which TRCs vary with both nanocluster and gas temperature, i.e. the manner in which the reemitted gas molecule energy from a surface varies with the temperature of that surface relative to that of the gas. In addition, while calculations ultimately reveal temperature insensitive TACs and justify their continued use in modeling gas-phase heat transfer, we argue that the TRC is more appropriate to calculate a priori in analyzing simulations; it is physically more realistic to first introduce a correction factor solely for the energy of reemitted gas molecules after a collision (as TRCs) instead of a correction factor applying to both the reemission energy and the initial energy (as TACs), as the latter is exactly calculable and independent of nanocluster properties. In line with prior work, for monoatomic gases we find the TAC decreases as the nanocluster to gas atomic mass increases, approaching 0.08 for gold nanoclusters in helium.

77 NANOSCIENCE AND NANOTECHNOLOGY↗

Accelerating Radiation Computations for Dynamical Models With Targeted Machine Learning and Code Optimization

Abstract Atmospheric radiation is the main driver of weather and climate, yet due to a complicated absorption spectrum, the precise treatment of radiative transfer in numerical weather and climate models is computationally unfeasible. Radiation parameterizations need to maximize computational efficiency as well as accuracy, and for predicting the future climate many greenhouse gases need to be included. In this work, neural networks (NNs) were developed to replace the gas optics computations in a modern radiation scheme (RTE+RRTMGP) by using carefully constructed models and training data. The NNs, implemented in Fortran and utilizing BLAS for batched inference, are faster by a factor of 1–6, depending on the software and hardware platforms. We combined the accelerated gas optics with a refactored radiative transfer solver, resulting in clear‐sky longwave (shortwave) fluxes being 3.5 (1.8) faster to compute on an Intel platform. The accuracy, evaluated with benchmark line‐by‐line computations across a large range of atmospheric conditions, is very similar to the original scheme with errors in heating rates and top‐of‐atmosphere radiative forcings typically below 0.1 K day −1 and 0.5 W m −2 , respectively. These results show that targeted machine learning, code restructuring techniques, and the use of numerical libraries can yield material gains in efficiency while retaining accuracy.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Integrated End-to-end Performance Prediction and Diagnosis for Extreme Scientific Workflows

This report details recent progress for the ASCR funded project “Integrated End-to-end Performance Prediction and Diagnosis for Extreme Scientific Workflows”. We refer to the project as IPPD/2, reflecting the 2017 renewal under expanded scope and partners In IPPD/2, we increased our research scope to include data motion. We are focusing on three major aspects: a) observe how data is generated, distributed, and used; b) analyze how data is (repeatedly) consumed with a focus both on repeated patterns and anomalies; and c) explore how to optimize data motion. This new work on data motion will augment and complement IPPD/2’s research that focused on the computational aspects of tasks. We leverage and extend our existing tools and demonstrate our work on the Belle II workflow suite as well as on workflows from NSLS-II. The highlights of our work are as follows: Provenance for Workflows: Provenance is used to provide information enabling quality control, re-run computational workflows, and reproduce results. IPPD/2 has been building a scalable provenance management system that enables the capture of provenance from the high-level workflow through all relevant system levels in one integrated environment. Leveraging this work, our recent efforts have included using provenance as an enabling technique. Workload characterization: Leveraging provenance and analysis, we characterize data movement within network, storage, and memory over a variety of workloads. This characterization enables an understanding by performance analysts and application developers of the range of behaviors that could be expected. Performance Prediction for Workflows: The goal of modeling distributed workflows is to understand performance bottlenecks and enable more intelligent task scheduling to optimize selected metrics of interest (e.g., task throughput or output data rate). IPPD/2 has utilized both analytical and AI/ML modeling methodologies for performance modeling. Advanced Scheduling and Fault Modeling for Workflows: Scheduling of large-scale scientific workflows on geographically distributed resources is a challenging problem. To improve workflow throughput, we combined novel scheduling algorithms with task predictions from performance modeling and fault modeling. Dynamically Alleviating Bottlenecks in Workflows: Exploiting our provenance, analysis, and modeling efforts, we have explored and developed several techniques for dynamically detecting and alleviating bottlenecks in data movement. In particular, we have spent considerable effort demonstrating our techniques on production-like workflow configurations.

97 MATHEMATICS AND COMPUTING↗