Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “MODEL SIMULATION DATA”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 1,027 records · Page 57

Plasma Reactor Modeling and Validation Experiments

Plasma processing is a key processing stop in integrated circuit manufacturing. Low pressure, high density plum reactors are widely used for etching and deposition. Inductively coupled plasma (ICP) source has become popular recently in many processing applications. In order to accelerate equipment and process design, an understanding of the physics and chemistry, particularly, plasma power coupling, plasma and processing uniformity and mechanism is important. This understanding is facilitated by comprehensive modeling and simulation as well as plasma diagnostics to provide the necessary data for model validation which are addressed in this presentation. We have developed a complete code for simulating an ICP reactor and the model consists of transport of electrons, ions, and neutrals, Poisson's equation, and Maxwell's equation along with gas flow and energy equations. Results will be presented for chlorine and fluorocarbon plasmas and compared with data from Langmuir probe, mass spectrometry and FTIR.

Meyyappan, M.↗

Comparative Analysis of ANN and LSTM Prediction Accuracy and Cooling Energy Savings through AHU-DAT Control in an Office Building

This paper proposes the optimal algorithm for controlling the HVAC system in the target building. Previous studies have analyzed pre-selected algorithms without considering the unique data characteristics of the target building, such as location, climate conditions, and HVAC system type. To address this, we compare the accuracy of cooling load prediction using ANN and LSTM algorithms, widely used in building energy research, to determine the optimal algorithm for HVAC control in the target building. We develop a simulation model calibrated with actual data to ensure data reliability and compare the energy consumption of the existing HVAC control method and the two algorithms-based methods. Results show that the ANN algorithm, with a CV(RMSE) of 12.7%, has a higher prediction accuracy than the LSTM algorithm, CV(RMSE) of 17.3%, making it a more suitable algorithm for HVAC control. Furthermore, implementing the ANN-based approach results in a 3.2% cooling energy reduction from the optimal control of Air Handling Unit (AHU) Discharge Air Temperature (DAT) compared to the fixed DAT at 12.8 °C in a representative day. This study demonstrates that ML-based HVAC system control can effectively reduce cooling energy consumption in HVAC systems, providing an effective strategy for energy conservation and improved HVAC system efficiency.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Development of machine learning framework for interface force closures based on bubble tracking data

Interfacial force closures in the two-fluid model play a critical role for the predictive capabilities of void fraction distribution. However, the practices of interfacial force modeling have long been challenged by the inherent physical complexity of the two-phase flows. The rapidly expanding computational capabilities in the recent years have made high-fidelity data from the interface-captured direct numerical simulation become more available, and hence potential for data-driven interfacial force modeling has prevailed. In this work, we established a data-driven modeling framework integrated to the HZDR multiphase Eulerian-Eulerian framework for computational fluid dynamics simulations. The data-driven framework is verified in a benchmark problem, where a feedforward neural network managed to capture the non-linear mapping between bubble Reynolds number and drag coefficient and reproduce the void distribution resulting from the baseline model in the test case. The second focus is on utilizing the bubble tracking data set to form a closure for the bubble drag in the turbulent bubbly flow, in which the drag coefficient is set to be correlated with the bubble Reynolds number and the Eötvös number. Pseudo-steady state filtering in the Frenet Frame was carried out to obtain the drag coefficient from the turbulent bubbly flow data. The performance of the data-driven drag model is also examined through a case study, where improvement of model’s prediction near-wall is regarded necessary. In conclusion, discussion and further plans of investigation are provided.

42 ENGINEERING↗

Online Visualization and Analysis of Merged Global Geostationary Satellite Infrared Dataset

The NASA Goddard Earth Sciences Data Information Services Center (GES DISC) is home of Tropical Rainfall Measuring Mission (TRMM) data archive. The global merged IR product also known as the NCEP/CPC 4-km Global (60 degrees N - 60 degrees S) IR Dataset, is one of TRMM ancillary datasets. They are globally merged (60 degrees N - 60 degrees S) pixel-resolution (4 km) IR brightness temperature data (equivalent blackbody temperatures), merged from all available geostationary satellites (GOES-8/10, METEOSAT-7/5 and GMS). The availability of data from METEOSAT-5, which is located at 63E at the present time, yields a unique opportunity for total global (60 degrees N- 60 degrees S) coverage. The GES DISC has collected over 8 years of the data beginning from February of 2000. This high temporal resolution dataset can not only provide additional background information to TRMM and other satellite missions, but also allow observing a wide range of meteorological phenomena from space, such as, mesoscale convection systems, tropical cyclones, hurricanes, etc. The dataset can also be used to verify model simulations. Despite that the data can be downloaded via ftp, however, its large volume poses a challenge for many users. A single file occupies about 70 MB disk space and there is a total of approximately 73,000 files (approximately 4.5 TB) for the past 8 years. In order to facilitate data access, we have developed a web prototype to allow users to conduct online visualization and analysis of this dataset. With a web browser and few mouse clicks, users can have a full access to over 8 year and over 4.5 TB data and generate black and white IR imagery and animation without downloading any software and data. In short, you can make your own images! Basic functions include selection of area of interest, single imagery or animation, a time skip capability for different temporal resolution and image size. Users can save an animation as a file (animated gif) and import it in other presentation software, such as, Microsoft PowerPoint. The prototype will be integrated into GIOVANNI and existing GIOVANNI capabilities, such as, data download, Google Earth KMZ, etc will be available. Users will also be able to access other data products in the GIOVANNI family.

Liu, Zhong↗

A Non‐Intrusive Machine Learning Framework for Debiasing Long‐Time Coarse Resolution Climate Simulations and Quantifying Rare Events Statistics

Abstract Due to the rapidly changing climate, the frequency and severity of extreme weather is expected to increase over the coming decades. As fully‐resolved climate simulations remain computationally intractable, policy makers must rely on coarse‐models to quantify risk for extremes. However, coarse models suffer from inherent bias due to the ignored “sub‐grid” scales. We propose a framework to non‐intrusively debias coarse‐resolution climate predictions using neural‐network (NN) correction operators. Previous efforts have attempted to train such operators using loss functions that match statistics. However, this approach falls short with events that have longer return period than that of the training data, since the reference statistics have not converged. Here, the scope is to formulate a learning method that allows for correction of dynamics and quantification of extreme events with longer return period than the training data. The key obstacle is the chaotic nature of the underlying dynamics. To overcome this challenge, we introduce a dynamical systems approach where the correction operator is trained using reference data and a coarse model simulation nudged toward that reference. The method is demonstrated on debiasing an under‐resolved quasi‐geostrophic model and the Energy Exascale Earth System Model (E3SM). For the former, our method enables the quantification of events that have return period two orders longer than the training data. For the latter, when trained on 8 years of ERA5 data, our approach is able to correct the coarse E3SM output to closely reflect the 36‐year ERA5 statistics for all prognostic variables and significantly reduce their spatial biases.

Barthel Sorensen, B.↗

Comparison of simulation modeling and satellite techniques for monitoring ecological processes

In 1985 improvements were made in the world climatic data base for modeling and predictive mapping; in individual process models and the overall carbon-balance models; and in the interface software for mapping the simulation results. Statistical analysis of the data base was begun. In 1986 mapping was shifted to NASA-Goddard. The initial approach involving pattern comparisons was modified to a more statistical approach. A major accomplishment was the expansion and improvement of a global data base of measurements of biomass and primary production, to complement the simulation data. The main accomplishments during 1987 included: production of a master tape with all environmental and satellite data and model results for the 1600 sites; development of a complete mapping system used for the initial color maps comparing annual and monthly patterns of Normalized Difference Vegetation Index (NDVI), actual evapotranspiration, net primary productivity, gross primary productivity, and net ecosystem production; collection of more biosphere measurements for eventual improvement of the biological models; and development of some initial monthly models for primary productivity, based on satellite data.

Box, Elgene O.↗

Using the HARV simulation aerodynamic model to determine forebody strake aerodynamic coefficients from flight data

The method described in this report is intended to present an overview of a process developed to extract the forebody aerodynamic increments from flight tests. The process to determine the aerodynamic increments (rolling pitching, and yawing moments, Cl, Cm, Cn, respectively) for the forebody strake controllers added to the F/A - 18 High Alpha Research Vehicle (HARV) aircraft was developed to validate the forebody strake aerodynamic model used in simulation.

Messina, Michael D.↗

Generalizing to new geometries with Geometry-Aware Autoregressive Models (GAAMs) for fast calorimeter simulation

Generation of simulated detector response to collision products is crucial to data analysis in particle physics, but computationally very expensive. One subdetector, the calorimeter, dominates the computational time due to the high granularity of its cells and complexity of the interactions. Generative models can provide more rapid sample production, but currently require significant effort to optimize performance for specific detector geometries, often requiring many models to describe the varying cell sizes and arrangements, without the ability to generalize to other geometries. Here, we develop a geometry-aware autoregressive model, which learns how the calorimeter response varies with geometry, and is capable of generating simulated responses to unseen geometries without additional training. The geometry-aware model outperforms a baseline unaware model by over 50% in several metrics such as the Wasserstein distance between the generated and the true distributions of key quantities which summarize the simulated response. A single geometry-aware model could replace the hundreds of generative models currently designed for calorimeter simulation by physicists analyzing data collected at the Large Hadron Collider. This proof-of-concept study motivates the design of a foundational model that will be a crucial tool for the study of future detectors, dramatically reducing the large upfront investment usually needed to develop generative calorimeter models.

47 OTHER INSTRUMENTATION↗

A Bottom-up Approach to Improve the Representation of Deep Convective Clouds in Weather and Climate Models (Final Report)

A fundamental yet outstanding question about deep cumulus convection regards how convective updrafts, downdrafts, and surface-based cold pools are inter-related within organized storms, and how these three convective components are modulated by external and internal factors. We considered this question through numerical model simulations as well as data collected during the MC3E field campaign. One of our basic hypotheses was that deep cold pools are forced by wide downdrafts which arise from wide updrafts. This “bottom-up” perspective required consideration of the external controls on updraft width, and then the internal controls on cold-pool characteristics including depth. Model experiments showed that environmental vertical wind shear (an external factor) exerts a particularly strong control on updraft width; analyses of linear and nonlinear dynamic pressure forcing readily explain this control. The experiments also showed that increases in updraft width are accompanied by corresponding increases in downdraft width and cold-pool depth. The updraft–downdraft connection is through precipitation: the wide area of precipitation that will be initiated in a wide updraft will in turn lead to a wide downdraft. The downdraft–cold pool connection is through mass continuity: at the downdraft terminus, the large downdraft mass flux through a wide downdraft must lead to a deep cold pool. Model experiments also showed that the cold-pool depth and other characteristics can be significantly modulated by microphysical processes (internal factors). Specifically, the cold-pool expansion rate, depth, and intensity was found to depend most strongly on the amount of rain evaporation near the ground. However, the sustenance of the cold pool itself depended upon the downdrafts strengthened by sublimation of graupel higher in the storm. Finally, field campaign data revealed compelling inter-relationships between the cold-pool characteristics, and downdraft transience in terms of the microphysics. For example, pulses of hail and graupel (also internal factors) within downdrafts resulted in short-lived increases in the cold-pool depth. However, the cold-pool speed appeared to be relatively independent of the cold-pool depth and intensity, suggesting limitations of density current theory to convective cold pools in nature.

54 ENVIRONMENTAL SCIENCES↗

Rectification of single and multiple frames of satellite scanner imagery using points and edges as control

Rectification of single and overlapping multiple scanner frames produced by such satellite-borne scanners as the LANDSAT MSS was carried out using a newly developed comprehensive parametric model. Tests with both simulated and real image data demonstrate conclusively that this model in general is superior to the widely used polynomial model, and that the simultaneous rectification of overlapping frames using least squares techniques yields a high accuracy than sngle frame rectification due to the inclusion of tie points between the image frames. Used to control, edges or lines, whic are much more likely to be found in images, can replace conventional control points and can easily be implemented into the least squares approach. An efficient algorithm for findng corresponding points in image paris was developed which can be used for determining tie points between image frames and thus increase the ecnomy of the whole rectification procedure.

Paderes, F. C., Jr.↗

Simulating Complex Window Systems using BSDF Data

Nowadays, virtual models are commonly used to evaluate the performance of conventional window systems. Complex fenestration systems can be difficult to simulate accurately not only because of their geometry but also because of their optical properties that scatter light in an unpredictable manner. Bi-directional Scattering Distribution Functions (BSDF) have recently been developed based on a mixture of measurements and modelling to characterize the optics of such systems. This paper describes the workflow needed to create then use these BSDF datasets in the Radiance lighting simulation software. Limited comparisons are made between visualizations produced using the standard ray-tracing method, the BSDF method, and that taken in a full-scale outdoor mockup.

29 ENERGY PLANNING, POLICY, AND ECONOMY↗

Middle Atmosphere of the Southern Hemisphere (MASH) Global meteor observations system (GLOBMET) Solar Spectral Irradiance Measurements (SSIM) Global Observations and Studies of Stratospheric Aerosols (GOSSA): Progress with the MASH project

The aim of the MASH project is to study the dynamics of the middle atmosphere in the Southern Hemisphere, emphasizing inter-hemispheric differences. Both observational data and data from simulations with numerical models are being used. It is intended that MASH will be complemented by parallel studies on the transport and photochemistry of trace species in the Southern Hemisphere. Impetus for such studies has come from the unexpected finding of a springtime ozone hole over Antarctica. A summary of recent progress with the MASH project is given. Data from polar orbiting satellites are used to discuss the large scale circulation found in the Southern Hemisphere at extratropical latitudes. Comparisons are made with that of the Northern Hemisphere. Particular attention is paid to the springtime final warming, the most spectacular large scale phenomenon in the statosphere of the Southern Hemisphere. The circulation before and after this event has to be taken into account in theories for the formation and subsequent disappearance of the ozone hole.

Oneill, A.↗

Coupled Calculations in Helical Steam Generator: Validation on Legacy Data

The NEAMS program aims to develop an integrated multi-physics simulation capability “pellet-to-plant” for the design and analysis of future generations of nuclear power plants. In particular, the Reactor Product Line code suite's multi-resolution hierarchy is being designed to ultimately span the full range of length and time scales present in relevant reactor design and safety analyses, as well as scale from desktop to petaflop computing platforms. Flow-induced vibration (FIV) is widespread problem in energy systems because they rely on fluid movement for energy conversion. Vibrating structures may be damaged as fatigue or wear occurs. Given the importance of reliable components in the nuclear industry, flow-induced vibration has long been a major concern in safety and operation of nuclear reactors. In particular, nuclear fuel rods and steam generators have been known to suffer from flow-induced vibration and related failures. Advanced reactors, such as integral Pressurized Water Reactors (PWRs) considered for Small Modular Reactors (SMR), often rely on innovative component designs to meet cost and safety targets. One component that is the subject of advanced designs is the steam generator, some designs of which forego the usual shell-and-tube architecture in order to fit within the primary vessel. In addition to being more cost- and space-efficient, such steam generators need to be more reliable, since failure of the primary vessel represents a potential loss of coolant and a safety concern. A significant amount of data exists on flow-induced vibration in shell-and-tube heat exchangers, and heuristic methods are available to predict their occurrence based on a set of given assumptions. In contrast, advanced designs have far less data available. Advanced modeling and simulation based on coupled structural and fluid simulations have the potential to predict flow-induced vibration in a variety of designs, reducing the need for expensive experimental programs, especially at the design stage. Over the past five years, the Reactor Product Line has developed the integrated multi-physics code suite SHARP. The goal of developing such a tool is to perform multi-physics neutronics, thermal/fluid, and structural mechanics modeling of the components inside the full reactor core or portions of it with a user-specified fidelity. In particular SHARP contains high-fidelity single-physics codes Diablo for structural mechanics and Nek5000 for fluid mechanics calculations. Both codes are state-of-the-art, highly scalable tools that have been extensively validated. These tools form a strong basis on which to build a flow-induced vibration modeling capability. In this report we discuss one-way coupled calculations performed with Nek5000 and Diablo aimed at simulating available FIV experiments in helical steam generators in the turbulent buffeting regime. In this regime one-way coupling is judged sufficient because the pressure loads do not cause substantial displacements. It is also the most common source of vibration in helical steam generators at the low flows expected in integral PWRs. The legacy data is obtained from two datasets developed at Argonne and B&W.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗

Application of new simulation algorithms for modeling rf diagnostics of electron clouds

Traveling wave rf diagnostics of electron cloud build-up show promise as a non-destructive technique for measuring plasma density and the efficacy of mitigation techniques. However, it is very difficult to derive an absolute measure of plasma density from experimental measurements for a variety of technical reasons. Detailed numerical simulations are vital in order to understand experimental data, and have successfully modeled build-up. Such simulations are limited in their ability to reproduce experimental data due to the large separation of scales inherent to the problem. Namely, one must resolve both rf frequencies in the GHz range, as well as the plasma modulation frequency of tens of MHz, while running for very long simulations times, on the order of microseconds. The application of new numerical simulation techniques allow us to bridge the simulation scales in this problem and produce spectra that can be directly compared to experiments. The first method is to use a plasma dielectric model to measure plasma-induced phase shifts in the rf wave. The dielectric is modulated at a low frequency, simulating the effects of multiple bunch crossings. This allows simulations to be performed without kinetic particles representing the plasma, which both speeds up the simulations as well as reduces numerical noise from interpolation of particle charge and currents onto the computational grid. Secondly we utilize a port boundary condition model to simultaneously absorb rf at the simulation boundaries, and to launch the rf into the simulation. This method improves the accuracy of simulations by restricting rf frequencies better than adding an external (finite) current source to drive rf, and absorbing layers at the boundaries. We also explore the effects of non-uniform plasma densities on the simulated spectra.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Collaborative Research: Enabling multi-scale studies of magnetic reconnection with interpretable data-driven models

The development of accurate reduced descriptions and improved closures for magnetic reconnection is an important and a long‐standing challenge in plasma physics. The four‐fluid approach, and associated closures, that were investigated have the potential to improve the accuracy of plasma fluid models, capturing physical effects which would otherwise require a kinetic description. If successful, this approach could have an important impact for the modeling of laboratory and space plasmas. The major goals of this project were to develop new machine learning (ML) tools based on sparse and symbolic regression techniques, and to extract interpretable and generalizable reduced models (e.g., in the form of partial differential equations - PDEs) from data generated by first principles plasma simulations. Preserving interpretability of such data‐driven models is key to addressing the long‐standing theoretical and numerical challenges. Prior proof‐of‐principle studies have demonstrated the enormous potential of this approach, by recovering the well‐established hierarchy of plasma equations (from Vlasov to MHD) from data produced by particle‐in‐cell (PIC) simulations. Our goal in this project was to extend and apply these new tools to construct better kinetic closures for magnetic reconnection; to derive better models of particle injection and acceleration by this fundamental plasma process; and to use this understanding to accelerate the development of multi‐scale plasma algorithms. While our immediate focus was on the problem of magnetic reconnection, the tools that were will developed are general and applicable to other areas of plasma physics, and more broadly to many‐body phenomena. We anticipate that the development of these multi‐scale models will have a significant impact across different areas of plasma science, from fusion to space and astrophysical plasmas.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Understanding coarsening of a post-corrosion microstructure in a molten salt by combining phase-field modeling and in situ tomography

Alloys corroding in molten salt have been observed to form bicontinuous, nanoporous microstructures via dealloying, which subsequently undergo coarsening due to facile transport in high-temperature conditions. In this work, we describe a methodology to elucidate the underlying transport mechanisms during coarsening of a bicontinuous microstructure via quantitative comparisons between phase-field simulations and four-dimensional in situ experiments, in this case X-ray nanotomography of the coarsening of a dealloyed 80 wt% Ni-20 wt% Cr microwire in molten KCl-MgCl 2 at 800°C. We conduct phase-field simulations initialized from experimental data to model coarsening via three different transport mechanisms: surface diffusion, solid bulk diffusion, and liquid bulk diffusion. These simulations reproduce key features of the experiment, such as the densification of the outer layer of the dealloyed wire and the reduction in radius over time. We quantitatively compare different microstructural characteristics between the simulations and experiment and extract temporal scaling factors that optimally match the time scales of the simulations to that of the experiment. This allows us to evaluate morphological similarity between the simulations and experiment and relate the experimental coarsening kinetics to fundamental material properties. We find that surface diffusion is most likely to be the dominant coarsening mechanism, and its kinetics imply a surface diffusivity of D S = 8.9 x 10 -20 m 3 /s, which is within the range of reported values for Ni-vacuum interfaces at 800°C. However, the difference between the experiment and the surface diffusion simulation increases substantially at late times, suggesting that other mechanisms, such as the dissolution of residual Cr, may be at play.

36 MATERIALS SCIENCE↗

Influence of Nucleation Mechanisms on the Radiative Properties of Deep Convective Clouds and Subvisible Cirrus in CRYSTAL/FACE

During the past few years we have conducted work on several different topics, as reflected by our publications. As one of the Co-Project scientists for The Cirrus Regional Study of Tropical Anvils and Cirrus Layers - Florida Area Cirrus Experiment (CRYSTAL FACE) we worked to help design the mission and then conduct it in the field. Another major activity during the past two years has been to pull together various groups to formulate plans for follow on missions to CRYSTAL FACE. We organized a workshop at the University of Colorado during the summer of 2003 to assess the best locations for future missions. Working with a group of about 10 scientists from around the country we prepared a science-planning document (Tropical Composition, Cloud and Climate Coupling Experiment (TC(sup 4)) that outlined the rationale, locations, strategy to accomplish the goals, and possible payloads for a set of three tropical missions. We also prepared background materials for various NRAs being prepared at NASA Headquarters for missions in Costa Rica, Darwin and Guam. In conjunction with the group at NASA Ames we have helped build a new numerical model for deep convection and have applied that model to simulate the CRYSTAL data. Our goal in particular has been to better understand how convection distributes water vapor isotopes. CRYSTAL observations of water isotopes are very different from those suggested by previous workers who assumed the isotopes would obey Rayleigh fractionation. The water isotope study has several implications. First it is a check on the realism of the deep convection model. Second, the isotopes are a measure of the precipitation removal in the atmosphere. Hence they provide a constraint on a parameter that is difficult to otherwise measure. Finally it has been suggested that isotopes may be the key to unraveling the water transport into the stratosphere and upper troposphere. Such transport is critical both for the radiation balance and for stratospheric chemistry. Ours is the first model that is able to treat this transport. Our initial results have just been submitted to Geophys. Res. Lett (Smith et al., 2005). Essentially we are able to explain the vertical profiles of isotopes in the tropical tropopause transition layer. We are also able to account for stratospheric humidity and isotope abundances with this model. We have also been heavily involved in trying to improve our understanding of nitric acid condensation on ice. Gao et al (2004) have shown that water supersaturations above ice occur when the atmosphere is supersaturated with respect to nitric acid trihydrate. As one of the co-authors of that work, we suggested the mechanism that may explain why this is occurring. Essentially, ice does not like to grow near unit supersaturation, but does so because the water molecules can find sites on the ice surface to attach themselves to before they fly off the ice surface. This phenomena was well known in the 1960s when it was a source of debate about whether condensation and evaporation coefficients for ice would be the same. Evaporation does not require any molecular orientation, while condensation does, so it was possible that the coefficients would differ. They don't differ because the water molecules rapidly move across the surface and find places to attach. Nitric acid may be occupying these preferred sites and therefore the water molecules can't find a desirable place to attach. We anticipate that this research will be the subject of laboratory work during the coming few years. Another possibility that has been suggested is that cubic ice is forming in clouds. We have measured the vapor pressure of cubic ice, and plan to publish that result in the next few months. We have also been working on additional aspects of the condensation of nitric acid on ice. With Y. Kondo we studied the condensation of NOy on ice using the SOLVE data. Gamblin et al. have continued this work. The CRYSTAL NOy and HNO, groups have shown th their data can be fit using standard Langmuir isotherms as suggested in some, but not all, laboratory studies. We have found in the SOLVE data set that this is not the case. Moreover some laboratory studies show there are important kinetic effects that may be occurring in the atmosphere limiting the transfer of nitric acid to the ice. The SOLVE data seem consistent with these studies. We are currently re-analyzing the CRYSTAL data to look for these kinetic effects. There are a number of implications of these studies. One of the more interesting is that the nitric acid coating on ice can be used as a cloud clock to determine how long the cloud parcel has been in existence. We have also been involved with several laboratory studies. We have worked to improve the database on ice optical constants, which are critical for remote sensing. We have also studied the ways in which ice nucleates on clays. We suspect now that the standard theories used for depositional ice nucleation are completely incorrect. Further work will be needed to develop a new theory.

Toon, Owen B.↗

Predicting Small Molecule Transfer Free Energies by Combining Molecular Dynamics Simulations and Deep Learning

Accurately predicting small molecule partitioning and hydrophobicity is critical in the drug discovery process. There are many heterogeneous chemical environments within a cell and entire human body. For example, drugs must be able to cross the hydrophobic cellular membrane to reach their intracellular targets, and hydrophobicity is an important driving force for drug–protein binding. Atomistic molecular dynamics (MD) simulations are routinely used to calculate free energies of small molecules binding to proteins, crossing lipid membranes, and solvation but are computationally expensive. Machine learning (ML) and empirical methods are also used throughout drug discovery but rely on experimental data, limiting the domain of applicability. We present atomistic MD simulations calculating 15,000 small molecule free energies of transfer from water to cyclohexane. This large data set is used to train ML models that predict the free energies of transfer. We show that a spatial graph neural network model achieves the highest accuracy, followed closely by a 3D-convolutional neural network, and shallow learning based on the chemical fingerprint is significantly less accurate. A mean absolute error of ~4 kJ/mol compared to the MD calculations was achieved for our best ML model. We also show that including data from the MD simulation improves the predictions, tests the transferability of each model to a diverse set of molecules, and show multitask learning improves the predictions. This work provides insight into the hydrophobicity of small molecules and ML cheminformatics modeling, and our data set will be useful for designing and testing future ML cheminformatics methods.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗