Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “three phase modeling”

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 1,729 records · Page 96

The thermopower in the temperature range T(sub c)-1000K and the bank spectrum of Bi-based superconductors

The temperature dependencies of thermopower, S, in the range T = T(sub c)-1000K as well as of resistivity and Hall coefficient in the range T = T(sub c)-300K for the single-phase ceramic samples Bi2Sr2Ca(1-x)Nd(x)Cu2O(y) have been measured. It was found that the S(T) dependencies in normal phase have three characteristic regions. Despite the fact that the S(T) dependencies in Bi-based high-T(sub c) superconductors (HTSC) differ essentially from ones in Y-based HTSC at T = T(sub c)-300K, the main feature of theirs (S(T) = const at high temperatures) retains in samples investigated at T is greater than 620K. The results obtained have been analyzed on the basis of the narrow-band model with the use of assumption of slight asymmetry of the conductive band. The band spectrum parameters of the samples studied have been calculated. An analysis of the tendencies in these parameters changes with samples composition varying enables to make the conclusion about the similarity of the main features of the conductive band structure in Y- and Bi-based HTSC.

Gasumyants, V. E.↗

Annual Research Briefs, 1998

The topics contained in this progress report are direct numerical simulation of turbulent non-premixed combustion with realistic chemistry; LES of non-premixed turbulent reacting flows with conditional source term estimation; measurements of the three-dimensional scalar dissipation rate in gas-phase planar turbulent jets; direct simulation of a jet diffusion flame; on the use of interpolating wavelets in the direct numerical simulation of combustion; on the use of a dynamically adaptive wavelet collocation algorithm in DNS (direct numerical simulation) of non-premixed turbulent combustion; 2D simulations of Hall thrusters; computation of trailing-edge noise at low mach number using LES and acoustic analogy; weakly nonlinear modeling of the early stages of bypass transition; interactions between freestream turbulence and boundary layers; interfaces at the outer boundaries of turbulent motions; largest scales of turbulent wall flows; the instability of streaks in near-wall turbulence; an implementation of the v(sup 2) - f model with application to transonic flows; heat transfer predictions in cavities; a structure-based model with stropholysis effects; modeling a confined swirling coaxial jet; subgrid-scale models based on incremental unknowns for large eddy simulations; subgrid scale modeling taking the numerical error into consideration; towards a near-wall model for LES of a separated diffuser flow; on the feasibility of merging LES with RANS (Reynolds Averaging Numerical simulation) for the near-wall region of attached turbulent flows; large-eddy simulation of a separated boundary layer; numerical study of a channel flow with variable properties; on the construction of high order finite difference schemes on non-uniform meshes with good conservation properties; development of immersed boundary methods for complex geometries; and particle methods for micro and macroscale flow simulations.

Spinks, Debra↗

Higher Order Modulation Intersymbol Interference Caused by Traveling-wave Tube Amplifiers

For the first time, a time-dependent, physics-based computational model has been used to provide a direct description of the effects of the traveling wave tube amplifier (TWTA) on modulated digital signals. The TWT model comprehensively takes into account the effects of frequency dependent AM/AM and AM/PM conversion; gain and phase ripple; drive-induced oscillations; harmonic generation; intermodulation products; and backward waves, Thus, signal integrity can be investigated in the presence of these sources of potential distortion as a function of the physical geometry and operating characteristics of the high power amplifier and the operational digital signal. This method promises superior predictive fidelity compared to methods using TWT models based on swept-amplitude and/or swept-frequency data. First, the TWT model using the three dimensional (3D) electromagnetic code MAFIA is presented. Then, this comprehensive model is used to investigate approximations made in conventional TWT black-box models used in communication system level simulations, To quantitatively demonstrate the effects these approximations have on digital signal performance predictions, including intersymbol interference (ISI), the MAFIA results are compared to the system level analysis tool, Signal Processing, Workstation (SPW), using high order modulation schemes including 16 and 64-QAM.

Kory, Carol L.↗

Acoustic Modeling and Analysis for the Space Shuttle Main Propulsion System Liner Crack Investigation

Cracks were found on bellows flow liners in the liquid hydrogen feedlines of several space shuttle orbiters in 2002. An effort to characterize the fluid environment upstream of the space shuttle main engine low-pressure fuel pump was undertaken to help identify the cause of the cracks and also provide quantitative environments and loads of the region. Part of this effort was to determine the duct acoustics several inches upstream of the low-pressure fuel pump in the region of a bellows joint. A finite element model of the complicated geometry was made using three-dimensional fluid elements. The model was used to describe acoustics in the complex geometry and played an important role in the investigation. Acoustic mode shapes and natural frequencies of the liquid hydrogen in the duct and in the cavity behind the flow liner were determined. Forced response results were generated also by applying an edgetone-like forcing to the liner slots. Studies were conducted for state conditions and also conditions assuming two-phase entrapment in the backing cavity. Highly instrumented single-engine hot fire data confirms the presence of some of the predicted acoustic modes.

Casiano, Matthew J.↗

NASA's PEM Fuel Cell Power Plant Development Program for Space Applications

NASA embarked on a PEM fuel cell power plant development program beginning in 2001. This five-year program was conducted by a three-center NASA team of Glenn Research Center (lead), Johnson Space Center, and Kennedy Space Center. The program initially was aimed at developing hardware for a Reusable Launch Vehicle (RLV) application, but more recently had shifted to applications supporting the NASA Exploration Program. The first phase of the development effort, to develop breadboard hardware in the 1-5 kW power range, was conducted by two competing vendors. The second phase of the effort, to develop Engineering Model hardware at the 10 kW power level, was conducted by the winning vendor from the first phase of the effort. Both breadboard units and the single engineering model power plant were delivered to NASA for independent testing. This poster presentation will present a summary of both phases of the development effort, along with a discussion of test results of the PEM fuel cell engineering model under simulated mission conditions.

Hoberecht, Mark↗

New Global Electron Density Observations from GPS-RO in the D- and E-Region Ionosphere

A novel retrieval technique is developed for electron density (N(sub e)) in the D- and E-region (80-120 km) using the high-quality 50-Hz GPS radio occultation (GPS-RO) phase measurements. The new algorithm assumes a slow, linear variation in the F-region background when the GPS-RO passes through the D- and E-region, and extracts the N(sub e) profiles at 80-130 km from the phase advance signal caused by N(sub e). Unlike the conventional Abel function, the new approach produces a sharp N(sub e) weighting function in the lower ionosphere, and the N(sub e) retrievals are in good agreement with the IRI (International Reference Ionosphere) model in terms of monthly maps, zonal means and diurnal variations. The daytime GPS-RO N(sub e) profiles can be well characterized by the alpha-Chapman function of three parameters (N(sub mE), h(sub mE) and H), showing that the bottom of E-region is deepening and sharpening towards the summer pole. At high latitudes the monthly GPS-RO N(sub e) maps at 80-120 km reveal clear enhancement in the auroral zones, more prominent at night, as a result of energetic electron precipitation (EEP) from the outer radiation belt. The D-/E-region auroral N(sub e) is strongly correlated with K(sub p) on a daily basis. The new N(sub e) data allow further comprehensive analyses of the sporadic E (E(sub s)) phenomena in connection with the background N(sub e) in the E-region. The layered (2-10 km) and fluctuated (less than 2 km) E(sub s) components, namely N(sub e_Layer) than N(sub e_Pert), are extracted with respect to the background N( sub e_Region) on a profile-by-profile basis. The N(sub e_Layer) component has a strong but highly-refined peak at approximately 105 km, with an amplitude smaller than N(sub e_Region) approximately by an order of magnitude. The N(sub e_Pert) component, which was studied extensively in the past, is approximately 2 orders of magnitude weaker than N(sub e_Layer). Both N(sub e_Layer) and N(sub e_Pert) are subject to significant diurnal and semidiurnal variations, showing downward progression with local time in amplitude. The 11-year solar cycle dominates the N(sub e) interannual variations, showing larger N(sub e_Region) and N(sub e_Layer) but smaller N(sub e_Pert) amplitudes in the solar maximum years. Enhanced Ne profiles are often observed in the polar winter, showing good correlation with solar proton events (SPEs) and geomagnetic activity. The new methodology offers great potential for retrieving low N(sub e) in the D-region, where radio propagation and communication blackouts can occur due to enhanced ionization. For space weather applications it is recommended for GPSRO operations to raise the top of high-rate data acquisition to approximately 140 km in the future.

GSP-R↗

The NANOGrav 12.5 yr Data Set: The Frequency Dependence of Pulse Jitter in Precision Millisecond Pulsars

Low-frequency gravitational-wave experiments require the highest timing precision from an array of the moststable millisecond pulsars. Several known sources of noise on short timescales in single radio pulsar observationsare well described by a simple model of three components: template fitting from a finite signal-to-noise ratio, pulsephase/amplitude jitter from single-pulse stochasticity, and scintillation errors from short-timescale interstellarscattering variations. Currently template-fitting errors dominate, but as radio telescopes push toward higher signalto-noise ratios, jitter becomes the next dominant term for most millisecond pulsars. Understanding the statistics ofjitter becomes crucial for properly characterizing arrival time uncertainties. We characterize the radio frequencydependence of jitter using data on 48 pulsars in the North American Nanohertz Observatory for GravitationalWaves timing program. We detect significant jitter in 43 of the pulsars and test several functional forms for itsfrequency dependence; we find significant frequency dependence for 30 pulsars. We find moderate correlations ofrms jitter with pulse width (R = 0.62) and number of profile components (R = 0.40); the single-pulse rms jitter istypically 1% of pulse phase. The average frequency dependence for all pulsars using a power-law model hasindex 0.42. We investigate the jitter variations for the interpulse of PSR B1937+21 and find no significantdeviations from the main pulse rms jitter. We also test the time variation of jitter in two pulsars and find thatsystematics likely bias the results for high-precision pulsars. Pulsar timing array analyses must properly modeljitter as a significant component of the noise within the detector.

Lam, M. T.↗

Dynamics of Sheared Granular Materials

This work focuses on the properties of sheared granular materials near the jamming transition. The project currently involves two aspects. The first of these is an experiment that is a prototype for a planned ISS (International Space Station) flight. The second is discrete element simulations (DES) that can give insight into the behavior one might expect in a reduced-g environment. The experimental arrangement consists of an annular channel that contains the granular material. One surface, say the upper surface, rotates so as to shear the material contained in the annulus. The lower surface controls the mean density/mean stress on the sample through an actuator or other control system. A novel feature under development is the ability to 'thermalize' the layer, i.e. create a larger amount of random motion in the material, by using the actuating system to provide vibrations as well control the mean volume of the annulus. The stress states of the system are determined by transducers on the non-rotating wall. These measure both shear and normal components of the stress on different size scales. Here, the idea is to characterize the system as the density varies through values spanning dense almost solid to relatively mobile granular states. This transition regime encompasses the regime usually thought of as the glass transition, and/or the jamming transition. Motivation for this experiment springs from ideas of a granular glass transition, a related jamming transition, and from recent experiments. In particular, we note recent experiments carried out by our group to characterize this type of transition and also to demonstrate/ characterize fluctuations in slowly sheared systems. These experiments give key insights into what one might expect in near-zero g. In particular, they show that the compressibility of granular systems diverges at a transition or critical point. It is this divergence, coupled to gravity, that makes it extremely difficult if not impossible to characterize the transition region in an earth-bound experiment. In the DE modeling, we analyze dynamics of a sheared granular system in Couette geometry in two (2D) and three (3D) space dimensions. Here, the idea is to both better understand what we might encounter in a reduced-g environment, and at a deeper level to deduce the physics of sheared systems in a density regime that has not been addressed by past experiments or simulations. One aspect of the simulations addresses sheared 2D system in zero-g environment. For low volume fractions, the expected dynamics of this type of system is relatively well understood. However, as the volume fraction is increased, the system undergoes a phase transition, as explained above. The DES concentrate on the evolution of the system as the solid volume fraction is slowly increased, and in particular on the behavior of very dense systems. For these configurations, the simulations show that polydispersity of the sheared particles is a crucial factor that determines the system response. Figures 1 and 2 below, that present the total force on each grain, show that even relatively small (10 %) nonuniformity of the size of the grains (expected in typical experiments) may lead to significant modifications of the system properties, such as velocity profiles, temperature, force propagation, and formation shear bands. The simulations are extended in a few other directions, in order to provide additional insight to the experimental system analyzed above. In one direction, both gravity, and driving due to vibrations are included. These simulations allow for predictions on the driving regime that is required in the experiments in order to analyze the jamming transition. Furthermore, direct comparison of experiments and DES will allow for verification of the modeling assumptions. We have also extended our modeling efforts to 3D. The (preliminary) results of these simulations of an annular system in zero-g environment will conclude the presentation.

Kondic, Lou↗

Relating phase transitions to pore size distributions and mechanical mortar properties in CSA-OPC-C$ based systems – The potential impact of delayed straetlingite formation

The scope of this research is to investigate the performance evolution of five ternary CSA-OPC-Cs systems. Time resolved thermodynamic modeling was employed to gain insight into changes in the volume fractions of the hydrating pastes and link these changes to trends in the pore size distributions of the mortars. Pore solution analyses were performed to investigate which hydrates have the potential to exert high crystallization pressures. In three out of five blends the resulting tensile stress exceeded the tensile strength of the materials, resulting in the occurrence of coarse pores, which could be linked to the measured compressive strengths. Two damaging crystallization processes were identified in the investigated blends. One is the continuous precipitation of ettringite at high supersaturation in a system with high anhydrite addition. The other is the delayed crystallization of straetlingite at high supersaturation and crystallization pressure in the confined environment of mature pastes.

36 MATERIALS SCIENCE↗

Feasibility study of a bidirectional jet flap device for application to helicopter rotor blades. Phase 2: Lift controller development

A bidirectional jet flap device called the variable deflection thruster (VDT) has been investigated for possible application to helicopter rotors. This investigation included the development and testing of a fluidic lift control system for the VDT-blade model making use of the test result that VDT-blade lift can be sensed from the differential pressure at midchord. This study constitutes a long-range program to develop blown control techniques for stabilizing the higher harmonic modes of helicopter rotors. Wind tunnel tests were conducted using a three-sectioned, two-dimensional VDT-blade model having individually controlled VDT jet flaps in each section. Steady-state tests were conducted without the fluidic lift controller (open loop) for both full-span blowing and for the model center section blowing only. Steady-state tests were conducted with the center section blowing only using the fluidic lift controller (close-loop) to control the lift on the model center section. Dynamic tests were conducted using the complete model with the VDT jet in the model center section oscillating at various frequencies and also using the model center section alone on a single endplate to obtain finite-aspect-ratio effects. Fair agreement was obtained between theory and experimental results.

Rose, R. E.↗

Three-Dimensional In Situ Observations of Polycrystalline Microstructure Evolution During Directional Solidification of Transparent Alloys Aboard the ISS and Quantitative Comparison with Numerical Modeling

Cellular/dendritic microstructures formed during solidification have a crucial influence on the mechanical properties of a wide range of structural alloys. By minimizing the amount of gravity-induced convection in the liquid, directional solidification experiments using transparent organic alloys conducted in the DECLIC-DSI onboard the International Space Station have provided unique 3D in situ observations of the spatiotemporal evolution of the solid-liquid interface during the formation of cellular and dendritic microstructures under purely diffusive growth conditions. Those observations have made it possible to perform benchmark quantitative comparisons with the predictions of state-of-the-art phase-field simulations of microstructure formation in 3D on experimentally relevant length and time scales. This talk will report quantitative comparisons between microgravity experiments and phase-field simulations in succinonitrile-camphor alloys of two different compositions, which shed new light on the selection of dendritic array structures, and on roles of macroscopic curvatures and subgrain boundaries between grains with a small misorientation with respect to the temperature gradient in the spatiotemporal evolution of the primary cellular spacing.

Kaihua Ji↗

Application of GOTHIC to Groundwater Transport Analysis - 20152

Migration of pollutants and hazardous wastes, potentially containing radioactive isotopes, via groundwater transport is a concern at most waste cleanup sites. Predictive analysis can be used to evaluate mitigating actions intended to minimize impact on the environment and public exposure. GOTHIC is a multipurpose thermal hydraulics code that is used extensively in the nuclear industry for design, licensing and operation evaluations. It combines the capabilities of typical one-dimensional system codes and the essential features of Computational Fluid Dynamics (CFD) codes for three-dimensional analysis. There are other codes that are specifically developed for groundwater transport analysis and the results presented here are consistent with prior analyses. However, GOTHIC has some unique features that offer advantages for applications related to nuclear waste. Most importantly, it has been developed and maintained under a Quality Assurance program in compliance with the requirements of 10CFR50 Appendix B [2] and applicable portions of ASME NQA-1 [3] since 1995. Available GOTHIC capabilities that make the code especially useful for groundwater transport of nuclear materials include: - Tracking of any number of tracer elements for contaminants and other species of interest; - Radioactive decay and progeny of tracer elements; - Adsorption/desorption of tracer elements; - Tracking of any number of dissolved gases; - Release and absorption for dissolved gases; - Vapor phase tracking; - Non-Newtonian fluid modeling. The general porous body modeling approach makes GOTHIC well suited to groundwater transport analysis. The multi-region modeling approach used by GOTHIC simplifies model construction for regions of varying hydrologic characteristics and focuses the computational effort on regions of particular interest while simultaneously capturing the macroscopic response and any feedback effects across the larger domain. The applicability of GOTHIC to groundwater transport applications is demonstrated by comparing code results with available analytic or semi-analytic solutions for groundwater behavior. (authors)

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

Quasi-Steady Simulation of Glaze Ice Accretion and Heat Transfer in the Supercooled Large Droplet Regime in Atmospheric Turbulent Air Flow

Abstract To avoid the use of a computationally intensive full unsteady simulation while providing an accurate solution, a quasi-steady simulation has been performed to study glaze and mixed ice accretion in the supercooled large droplet (SLD) regime in turbulent flow of atmospheric air. We have attempted to find the minimum time-step necessary to adequately simulate the icing process on an airfoil surface. Based on node displacement, a mesh morphing scheme has been adopted in the computations to account for the moving boundaries that are caused by the continuous ice buildup. We have modeled ice accretion on an airfoil surface for a time period of 232 s using several time steps. At each time-step we solved the steady-state conservation equations for the air and droplet phases and then used the results as initial conditions for the next time position. The magnitude of the time-step ranged from a least accurate two-shot simulation (where the time-step is 116 s) to a most accurate 2320-shot simulation (where the time-step is 0.1 s). In-between these two extreme time steps, we have performed a three-shot simulation (where the time-step is 77.33 s), a four-shot simulation (where the time-step is 58 s), a six-shot simulation (where the time-step is 38.67 s), a 46-shot simulation (where the time-step is 5 s), and a 232-shot simulation (where the time-step is 1 s). We have done so to find out the degree of accuracy (or inaccuracy) of the multishot simulation approach and to find out the appropriate time-step needed for a successful and valid quasi-steady simulation. A valid quasi-steady simulation needs to use a time-step that is small enough to reproduce the full time-dependent solution within a very small error band. We have found that both the 1 s and the 0.1 s time steps produce virtually identical results. This is the primary litmus test that proves the validity of the quasi-steady-state assumption. The results adopted in this paper are thus all based on the more conservative 0.1 s time-step. In the process of performing the simulations, remeshing was required in order to maintain the grid density in zones of high curvature to be able to capture the full physics in those zones. After successfully modeling glaze ice accretion over the airfoil surface using the 0.1 s quasi-steady simulation approach, the effects of supercooled large droplets (SLDs) impacting the surface have been examined and presented in terms of the variation of the local collection efficiency, the water film thickness, and the heat transfer rate. Examination of the variation of the angle that the ice horn makes with the airfoil chord line demonstrated a 20% improvement in angle prediction when the time-step is reduced from 116 s to 0.1 s. The analysis also reveals a 12% or 8.5% increase in the maximum collection efficiency, βmax, depending on whether the value of the liquid water content (LWC) has been doubled from 0.5 g/m3 to 1 g/m3, or the value of the freestream velocity has been doubled from 75 m/s to 150 m/s, respectively. Because of the need to monitor the local collection efficiency and convective heat fluxes at each shot (regardless of the number of shots employed), the approach adopted here was found to be effective in successively and successfully reproducing the curvature of the glaze ice horn.

Engineering↗

A Generalized Framework for Service Restoration in a Resilient Power Distribution System

An electric power grid is one of the complex infrastructures, and because of its complex nature, there is simply no way that outages can be completely avoided. Thus, a modern society that depends on reliable electric supply requires a resilient electric system that can recover from disruptions while integrating emerging smart grid technologies. Here, this article presents a novel approach for service restoration in a modern power distribution system with controllable switches and distributed generation (DG) resources for any kind of outage. The proposed framework supports both the traditional service restoration using feeder reconfiguration and the grid-forming DG-assisted intentional islands that are dynamically sized using algorithms based on the fault scenario, available resources, and priority of loads. The problem is formulated as a mixed-integer linear program that incorporates critical system connectivity and operating constraints. Simulations are performed to demonstrate the effectiveness of the proposed approach using a large-scale four-feeder 1069-bus three-phase unbalanced distribution test system. It is demonstrated that the framework is effective in utilizing all available resources in quickly restoring the power supply to improve resiliency during extreme events and is scalable for a large-scale unbalanced power distribution system.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Three-dimensional coherent X-ray diffraction imaging via deep convolutional neural networks

Abstract As a critical component of coherent X-ray diffraction imaging (CDI), phase retrieval has been extensively applied in X-ray structural science to recover the 3D morphological information inside measured particles. Despite meeting all the oversampling requirements of Sayre and Shannon, current phase retrieval approaches still have trouble achieving a unique inversion of experimental data in the presence of noise. Here, we propose to overcome this limitation by incorporating a 3D Machine Learning (ML) model combining (optional) supervised learning with transfer learning. The trained ML model can rapidly provide an immediate result with high accuracy which could benefit real-time experiments, and the predicted result can be further refined with transfer learning. More significantly, the proposed ML model can be used without any prior training to learn the missing phases of an image based on minimization of an appropriate ‘loss function’ alone. We demonstrate significantly improved performance with experimental Bragg CDI data over traditional iterative phase retrieval algorithms.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

Machine learning modeling and model predictive control of a closed-circuit reverse osmosis system

Closed-circuit reverse osmosis (CCRO) offers a flexible and energy-efficient alternative to conventional reverse osmosis by operating in a semi-batch mode that recycles brine, enabling higher recovery rates and reduced specific energy consumption (SEC). However, developing accurate, system-level dynamic models for CCRO remains challenging due to its nonlinear, multi-phase operation and sensitivity to variable feed water conditions. Traditional modeling approaches, such as NARMAX (nonlinear autoregressive moving average with exogenous inputs), often struggle to generalize across varying inlet feed concentrations, necessitating frequent parameter re-estimation and limiting their utility for real-time control applications. To address these limitations, we developed a long short-term memory (LSTM) neural network model trained on an extensive experimental data set from a CCRO pilot plant. The model accepts three inputs, feed flow rate, recirculation flow rate, and initial feed conductivity, and predicts three key outputs: reject conductivity, feed pump power draw, and recirculation pump power draw. We validated the LSTM model against experimental data, demonstrating its ability to distinguish between different feed conductivities and adapt to variable flow rates. Subsequently, we incorporated the LSTM model within a nonlinear model predictive control (MPC) scheme and conducted closed-loop simulations to optimize the integrated SEC (iSEC). In conclusion, the results project up to a 6% reduction in iSEC by using MPC to optimize performance over the entire experiment duration, without requiring any random excitation for data collection or parameter re-estimation.

Desalination↗

Graphical Gaussian Process Regression Model for Aqueous Solvation Free Energy Prediction of Organic Molecules in Redox Flow Battery

The solvation free energy of organic molecules is a critical parameter in determining emergent properties such as solubility, liquid-phase equilibrium constants, and pKa and redox potentials in an organic redox flow battery. In this work, we present a machine learning (ML) model that can learn and predict the aqueous solvation free energy of an organic molecule using Gaussian process regression method based on a new molecular graph kernel. To investigate the performance of the ML model on electrostatic interaction, the nonpolar interaction contribution of solvent and the conformational entropy of solute in solvation free energy, three data sets with implicit or explicit water solvent models, and contribution of conformational entropy of solute are tested. We demonstrate that our ML model can predict the solvation free energy of molecules at chemical accuracy with a mean absolute error of less than 1 kcal/mol for subsets of the QM9 dataset and the Freesolv database. To solve the general data scarcity problem for a graph-based ML model, we propose a dimension reduction algorithm based on the distance between molecular graphs, which can be used to examine the diversity of the molecular data set. It provides a promising way to build a minimum training set to improve prediction for certain test sets where the space of molecular structures is predetermined.

25 ENERGY STORAGE↗