Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Predictive 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 883 records · Page 49
All sourcesNASA NTRSDOE OSTI
DOE OSTI2015

Discovery of optimal zeolites for challenging separations and chemical transformations using predictive materials modeling

Zeolites play numerous important roles in modern petroleum refineries and have the potential to advance the production of fuels and chemical feedstocks from renewable resources. The performance of a zeolite as separation medium and catalyst depends on its framework structure. To date, 213 framework types have been synthesized and >330000 thermodynamically accessible zeolite structures have been predicted. Hence, identification of optimal zeolites for a given application from the large pool of candidate structures is attractive for accelerating the pace of materials discovery. Here we identify, through a large-scale, multi-step computational screening process, promising zeolite structures for two energy-related applications: the purification of ethanol from fermentation broths and the hydroisomerization of alkanes with 18-30 carbon atoms encountered in petroleum refining. Finally, these results demonstrate that predictive modeling and data-driven science can now be applied to solve some of the most challenging separation problems involving highly non-ideal mixtures and highly articulated compounds.

36 MATERIALS SCIENCE↗
DOE OSTI2021

Slot-die-coating operability windows for polymer electrolyte membrane fuel cell cathode catalyst layers

Roll-to-roll (R2R) slot-die coating of polymer electrolyte membrane fuel cell (PEMFC) catalyst layers represents a scalable deposition method for producing 10–20 m 2 ·min -1 of catalyst-coated gas diffusion layers (GDLs). This high-throughput production technique will help lower the cost of PEMFC catalyst layers. The uniformity of the wet layer applied by slot die deposition is affected by process parameters such as substrate speed, vacuum pressure applied at the upstream meniscus, gap between the slot die lips and substrate, ink rheology, and other ink and substrate properties. The set of conditions for producing a defect-free coating with a dilute ink typically requires little to no upstream vacuum pressure, so suitable operating conditions can be found easily through trial and error and operator intuition. However, the higher viscosity of more concentrated inks dramatically shifts the range of settings that result in a homogeneous coating to higher vacuum levels, which are harder to find through hit or miss. A predictive model showing the range of operable conditions decreases material wastage inherent in experimentally searching for suitable parameters. In this study, the defect-free coating parameter window is explored experimentally and theoretically for two concentrations of PEFC cathode inks. Both a full capillary hydrodynamic model and a computationally cheaper viscocapillary model successfully predict the experimentally determined coating window within the experimental and model uncertainty limits for inks with 5.3 wt% and 12.0 wt% solids ink while maintaining the 0.1 mg Pt ·cm -2 Pt areal loading target. Here, this paper demonstrates a viable pathway for meeting the $30/kW net ultimate cost target of the United States Department of Energy (U.S. DOE) Hydrogen Fuel Cells Technologies Office (HFTO). The concentrated ink lowers the thermal energy and capital expenditure (CapEx) budget of the coating process by decreasing the amount of time, energy, and floorspace required for drying the coating.

25 ENERGY STORAGE↗
DOE OSTI2015

Development of a physiologically based pharmacokinetic model for assessment of human exposure to bisphenol A

A previously developed physiologically based pharmacokinetic (PBPK) model for bisphenol A (BPA) in adult rhesus monkeys was modified to characterize the pharmacokinetics of BPA and its phase II conjugates in adult humans following oral ingestion. Coupled with in vitro studies on BPA metabolism in the liver and the small intestine, the PBPK model was parameterized using oral pharmacokinetic data with deuterated-BPA (d{sub 6}-BPA) delivered in cookies to adult humans after overnight fasting. The availability of the serum concentration time course of unconjugated d{sub 6}-BPA offered direct empirical evidence for the calibration of BPA model parameters. The recalibrated PBPK adult human model for BPA was then evaluated against published human pharmacokinetic studies with BPA. A hypothesis of decreased oral uptake was needed to account for the reduced peak levels observed in adult humans, where d{sub 6}-BPA was delivered in soup and food was provided prior to BPA ingestion, suggesting the potential impact of dosing vehicles and/or fasting on BPA disposition. With the incorporation of Monte Carlo analysis, the recalibrated adult human model was used to address the inter-individual variability in the internal dose metrics of BPA for the U.S. general population. Model-predicted peak BPA serum levels were in the range of pM, with 95% of human variability falling within an order of magnitude. This recalibrated PBPK model for BPA in adult humans provides a scientific basis for assessing human exposure to BPA that can serve to minimize uncertainties incurred during extrapolations across doses and species. - Highlights: • A PBPK model predicts the kinetics of bisphenol A (BPA) in adult humans. • Serum concentrations of aglycone BPA are available for model calibration. • Model predicted peak BPA serum levels for adult humans were in the range of pM. • Model predicted 95% of human variability fell within an order of magnitude.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2026

Revisiting the validity of eddy viscosity models for predicting airflow over water waves

In this study, we revisit the validity of eddy viscosity models for predicting wave-induced airflow disturbances over ocean surface waves. We first derive a turbulence curvilinear model for the phase-averaged Navier–Stokes equations, extending the work of Cao, Deng & Shen (2020 J. Fluid Mech. 901, A27), by incorporating turbulence stress terms previously neglected in the linearised viscous curvilinear model. To verify our formulation, we perform a priori tests by numerically solving the model using mean wind and turbulence stress profiles from large-eddy simulations (LES) of airflow over waves across various wave ages. Results show that including turbulence stress terms improves wave-induced airflow predictions compared with the previous viscous curvilinear model. We further show that using a standard mixing-length eddy viscosity yields inaccurate predictions at certain wave ages, as it fails to capture wave-induced turbulence, which fundamentally differs from mean shear-driven turbulence. The LES data show that accurate representations of wave-induced stresses require a complex-valued eddy viscosity. The maximum magnitude of this eddy viscosity scales as ∼𝑢 𝜏 ⁢𝜁 𝑖𝑛𝑛𝑒𝑟 , where 𝑢 𝜏 is the friction velocity and 𝜁 𝑖𝑛𝑛𝑒𝑟 is the inner-layer thickness, the height at which the eddy-turnover time matches the wave advection time scale. This scaling aligns with the prediction by Belcher & Hunt (1993 J. Fluid Mech. 251, 109–148). Overall, the findings demonstrate that traditional eddy viscosity models are inadequate for capturing wave-induced turbulence. More sophisticated turbulence models are essential for the accurate prediction of airflow disturbances and form drag in wind–wave interaction models.

16 TIDAL AND WAVE POWER↗
DOE OSTI2016

Centipod WEC, Advanced Controls, WaveDyn Models

Archive containing the WaveDyn models used for analysis of the Centipod WEC with and without Advanced Controls, i.e. the WaveDyn files contain a baseline model which can be run alone, and a model predictive control (MPC) model which must be run when connected to the MPC controller. The controller dynamic link library (DLL) is also contained within this archive. The code archive for the MPC controller is linked in this submission.

16 TIDAL AND WAVE POWER↗
DOE OSTI2025

Mapping tree height in complex terrain of northern China using ultra-high-resolution images

Tree height is a key parameter for estimating forest biomass and carbon sequestration. In recent years, notable progress has been made in mapping tree height using satellite imagery. However, existing tree height products show low accuracy in mountainous and complex terrains, and few studies typically addressed tree height estimations in mountain areas. This study examines the Mentougou district of Beijing, China, characterized by complex terrain and mountainous landscapes. We analyzed two methods for estimating tree height: one using only spectral features and another combining spectral features with topographic factors (elevation, slope, aspect). We used 3-m resolution PlanetScope 8-band multispectral imagery, with 710 field-measured individual tree heights averaged to obtain 471 pixel-level tree height values as ground-truth, to develop tree height prediction models using eXtreme Gradient Boosting (XGBoost), Random Forest (RF), and Gradient Boosting Machine (GBM) models. The results show that the XGBoost model consistently presented the highest accuracy for both methods evaluated. Specifically, the XGBoost model that combined spectral data with elevation and slope variables with an R² of 0.75 and an RMSE of 2.69 m. Using the XGBoost model, we generated the tree height map for the Mentougou area at 3 m resolution, showing tree heights ranging from 0.5 to 30.4 m, and the model’s prediction error standard deviations ranged from 2.50 to 4.71 m, indicating reliable performance across varied terrain. Additionally, we compared and evaluated the global tree height products, identifying limitations in the accuracy within complex terrains. This study demonstrates the potential for accurately predicting tree heights by combining high-resolution multispectral satellites with a terrain factor modeling approach.

Complex terrain↗
DOE OSTI2017

Development of models for predicting Torsade de Pointes cardiac arrhythmias using perceptron neural networks

Background: Blockage of some ion channels and in particular, the hERG (human Ether-a’-go-go-Related Gene) cardiac potassium channel delays cardiac repolarization and can induce arrhythmia. In some cases it leads to a potentially life-threatening arrhythmia known as Torsade de Pointes (TdP). Therefore recognizing drugs with TdP risk is essential. Candidate drugs that are determined not to cause cardiac ion channel blockage are more likely to pass successfully through clinical phases II and III trials (and preclinical work) and not be withdrawn even later from the marketplace due to cardiotoxic effects. The objective of the present study is to develop an SAR (Structure-Activity Relationship) model that can be used as an early screen for torsadogenic (causing TdP arrhythmias) potential in drug candidates. The method is performed using descriptors comprised of atomic NMR chemical shifts (13C and 15N NMR) and corresponding interatomic distances which are combined into a 3D abstract space matrix. The method is called 3D-SDAR (3-dimensional spectral data-activity relationship) and can be interrogated to identify molecular features responsible for the activity, which can in turn yield simplified hERG toxicophores. A dataset of 55 hERG potassium channel inhibitors collected from Kramer et al. consisting of 32 drugs with TdP risk and 23 with no TdP risk was used for training the 3D-SDAR model. Results: An artificial neural network (ANN) with multilayer perceptron was used to define collinearities among the independent 3D-SDAR features. A composite model from 200 random iterations with 25% of the molecules in each case yielded the following figures of merit: training, 99.2%; internal test sets, 66.7%; external (blind validation) test set, 68. 4%. In the external test set, 70.3% of positive TdP drugs were correctly predicted. Moreover, toxicophores were generated from TdP drugs. Conclusion: A 3D-SDAR was successfully used to build a predictive model for drug-induced torsadogenic and non-torsadogenic drugs based on 55 compounds. The model was tested in 38 external drugs.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2013

Adaptive Sampling Algorithms for Probabilistic Risk Assessment of Nuclear Simulations

Nuclear simulations are often computationally expensive, time-consuming, and high-dimensional with respect to the number of input parameters. Thus exploring the space of all possible simulation outcomes is infeasible using finite computing resources. During simulation-based probabilistic risk analysis, it is important to discover the relationship between a potentially large number of input parameters and the output of a simulation using as few simulation trials as possible. This is a typical context for performing adaptive sampling where a few observations are obtained from the simulation, a surrogate model is built to represent the simulation space, and new samples are selected based on the model constructed. The surrogate model is then updated based on the simulation results of the sampled points. In this way, we attempt to gain the most information possible with a small number of carefully selected sampled points, limiting the number of expensive trials needed to understand features of the simulation space. We analyze the specific use case of identifying the limit surface, i.e., the boundaries in the simulation space between system failure and system success. In this study, we explore several techniques for adaptively sampling the parameter space in order to reconstruct the limit surface. We focus on several adaptive sampling schemes. First, we seek to learn a global model of the entire simulation space using prediction models or neighborhood graphs and extract the limit surface as an iso-surface of the global model. Second, we estimate the limit surface by sampling in the neighborhood of the current estimate based on topological segmentations obtained locally. Our techniques draw inspirations from topological structure known as the Morse-Smale complex. We highlight the advantages and disadvantages of using a global prediction model versus local topological view of the simulation space, comparing several different strategies for adaptive sampling in both contexts. One of the most interesting models we propose attempt to marry the two by obtaining a coarse global representation using prediction models, and a detailed local representation based on topology. Our methods are validated on several analytical test functions as well as a small nuclear simulation dataset modeled after a simplified Pressurized Water Reactor.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2018

Flux balance modeling to predict bacterial survival during pulsed-activity events

Abstract. Desert biological soil crusts (BSCs) are cyanobacteria-dominated surface soil microbial communities common to plant interspaces in arid environments. The capability to significantly dampen their metabolism allows them to exist for extended periods in a desiccated dormant state that is highly robust to environmental stresses. However, within minutes of wetting, metabolic functions reboot, maximizing activity during infrequent permissive periods. Microcoleus vaginatus, a primary producer within the crust ecosystem and an early colonizer, initiates crust formation by binding particles in the upper layer of soil via exopolysaccharides, making microbial dominated biological soil crusts highly dependent on the viability of this organism. Previous studies have suggested that biopolymers play a central role in the survival of this organism by powering resuscitation, rapidly forming compatible solutes, and fueling metabolic activity in dark, hydrated conditions. To elucidate the mechanism of this phenomenon and provide a basis for future modeling of BSCs, we developed a manually curated, genome-scale metabolic model of Microcoleus vaginatus (iNJ1153). To validate this model, gas chromatography–mass spectroscopy (GC–MS) and liquid chromatography–mass spectroscopy (LC–MS) were used to characterize the rate of biopolymer accumulation and depletion in in hydrated Microcoleus vaginatus under light and dark conditions. Constraint-based flux balance analysis showed agreement between model predictions and experimental reaction fluxes. A significant amount of consumed carbon and light energy is invested into storage molecules glycogen and polyphosphate, while β-polyhydroxybutyrate may function as a secondary resource. Pseudo-steady-state modeling suggests that glycogen, the primary carbon source with the fastest depletion rate, will be exhausted if M. vaginatus experiences dark wetting events 4 times longer than light wetting events.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2018

Mechanisms of radiation-induced segregation around He bubbles in a Fe-Cr-Ni crystal

We report that Fe-15Cr-15Ni (wt.%) austenitic stainless steel was irradiated with 200 keV 4 He + ions to a fluence of 5 × 10 16 /cm 2 at 500 °C. Radiation-induced segregation (RIS) at He bubbles in the irradiated steel was measured using Energy Dispersive X-ray Spectroscopy (EDS) coupled with Scanning Transmission Electron Microscopy (STEM). The RIS measurements exhibited an enrichment of Ni and a depletion of Cr and Fe at the bubbles. To determine the underlying mechanisms of the RIS, the prediction of the Marwick model was compared with the RIS measurements. The model with parametric input in accordance with the Inverse Kirkendall effect (IKE), i.e. a preferential flux of elements associated with vacancy flux, was inadequate in explaining the observed amount of segregation of Fe, Cr and Ni. More specifically, a considerably lower ratio of the concentration gradient between Cr and Fe, $\left(\frac{∇Cr}{∇Fe}\right)$, of 0.53 was obtained from the RIS measurements versus the model-predicted value of 2.3, indicating an overestimation of Cr depletion at the bubbles by the Marwick model. To explain this difference between the model prediction and the RIS measurements, in addition to the Cr depletion at the bubbles by IKE due to the fastest diffusion via vacancy flux, a preferential interstitial flux of Cr from the matrix to the bubbles is discussed as a possible concurrently operating mechanism.

36 MATERIALS SCIENCE↗
DOE OSTI2025

Deep Learning Approaches for Predicting the Surface Tension of Ionic Liquids

Ionic liquids (ILs) are a novel class of solvents that have attracted significant attention due to their unique and tunable properties. Among their physiochemical characteristics, surface tension plays a critical role in various industrial applications including electrolytes, heat transfer fluids, and separation processes. However, because of the exploratory nature of IL design and the vast combinatorial space of possible anion–cation pairs, the experimental determination of these properties is often impractical, being both time-consuming and costly. To overcome these challenges, computational approaches are increasingly employed to develop accurate predictive models that can accelerate IL discovery and design. In this study, we present two deep learning (DL) models for predicting the surface tension of ILs across a broad temperature range at a constant pressure. The models use simplified molecular input line entry system, SMILES, representations of ILs to extract molecular features as inputs. Both DL models demonstrate excellent agreement with experimental data, achieving an R 2 value of 0.990 and a root-mean-square error of 0.792 mN/m. In conclusion, these results offer valuable insights for the rapid screening and rational design of ILs with tailored surface tension values.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2015

Use of a Dual-Structure Constitutive Model for Predicting the Long-Term Behavior of an Expansive Clay Buffer in a Nuclear Waste Repository

Expansive soils are suitable as backfill and buffer materials in engineered barrier systems to isolate heat-generating nuclear waste in deep geological formations. The canisters containing nuclear waste would be placed in tunnels excavated at a depth of several hundred meters. The expansive soil should provide enough swelling capacity to support the tunnel walls, thereby reducing the impact of the excavation-damaged zone on the long-term mechanical and flow-barrier performance. In addition to their swelling capacity, expansive soils are characterized by accumulating irreversible strain on suction cycles and by effects of microstructural swelling on water permeability that for backfill or buffer materials can significantly delay the time it takes to reach full saturation. In order to simulate these characteristics of expansive soils, a dual-structure constitutive model that includes two porosity levels is necessary. The authors present the formulation of a dual-structure model and describe its implementation into a coupled fluid flow and geomechanical numerical simulator. The authors use the Barcelona Basic Model (BBM), which is an elastoplastic constitutive model for unsaturated soils, to model the macrostructure, and it is assumed that the strains of the microstructure, which are volumetric and elastic, induce plastic strain to the macrostructure. The authors tested and demonstrated the capabilities of the implemented dual-structure model by modeling and reproducing observed behavior in two laboratory tests of expansive clay. As observed in the experiments, the simulations yielded nonreversible strain accumulation with suction cycles and a decreasing swelling capacity with increasing confining stress. Finally, the authors modeled, for the first time using a dual-structure model, the long-term (100,000 years) performance of a generic heat-generating nuclear waste repository with waste emplacement in horizontal tunnels backfilled with expansive clay and hosted in a clay rock formation. The thermo-hydro-mechanical results of the dual-structure model were compared with those of the standard single-structure BBM. The main difference between the simulation results from the two models is that the dual-structure model predicted a time to fully saturate the expansive clay barrier on the order of thousands of years, whereas the standard single-structure BBM yielded a time on the order of tens of years. These examples show that a dual-structure model, such as the one presented here, is necessary to properly model the thermo-hydro-mechanical behavior of expansive soils.

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗
DOE OSTI2017

Steady-State Plant Model to Predict Hydroden Levels in Power Plant Components

The National Renewable Energy Laboratory (NREL) and Acciona Energy North America developed a full-plant steady-state computational model that estimates levels of hydrogen in parabolic trough power plant components. The model estimated dissolved hydrogen concentrations in the circulating heat transfer fluid (HTF), and corresponding partial pressures within each component. Additionally for collector field receivers, the model estimated hydrogen pressure in the receiver annuli. The model was developed to estimate long-term equilibrium hydrogen levels in power plant components, and to predict the benefit of hydrogen mitigation strategies for commercial power plants. Specifically, the model predicted reductions in hydrogen levels within the circulating HTF that result from purging hydrogen from the power plant expansion tanks at a specified target rate. Our model predicted hydrogen partial pressures from 8.3 mbar to 9.6 mbar in the power plant components when no mitigation treatment was employed at the expansion tanks. Hydrogen pressures in the receiver annuli were 8.3 to 8.4 mbar. When hydrogen partial pressure was reduced to 0.001 mbar in the expansion tanks, hydrogen pressures in the receiver annuli fell to a range of 0.001 mbar to 0.02 mbar. When hydrogen partial pressure was reduced to 0.3 mbar in the expansion tanks, hydrogen pressures in the receiver annuli fell to a range of 0.25 mbar to 0.28 mbar. Our results show that controlling hydrogen partial pressure in the expansion tanks allows us to reduce and maintain hydrogen pressures in the receiver annuli to any practical level.

14 SOLAR ENERGY↗
DOE OSTI2015

Systematic drug safety evaluation based on public genomic expression (Connectivity Map) data: Myocardial and infectious adverse reactions as application cases

Adverse drug reaction (ADR) is of great importance to both regulatory agencies and the pharmaceutical industry. Various techniques, such as quantitative structure–activity relationship (QSAR) and animal toxicology, are widely used to identify potential risks during the preclinical stage of drug development. Despite these efforts, drugs with safety liabilities can still pass through safety checkpoints and enter the market. This situation raises the concern that conventional chemical structure analysis and phenotypic screening are not sufficient to avoid all clinical adverse events. Genomic expression data following in vitro drug treatments characterize drug actions and thus have become widely used in drug repositioning. In the present study, we explored prediction of ADRs based on the drug-induced gene-expression profiles from cultured human cells in the Connectivity Map (CMap) database. The results showed that drugs inducing comparable ADRs generally lead to similar CMap expression profiles. Based on such ADR-gene expression association, we established prediction models for various ADRs, including severe myocardial and infectious events. Drugs with FDA boxed warnings of safety liability were effectively identified. We therefore suggest that drug-induced gene expression change, in combination with effective computational methods, may provide a new dimension of information to facilitate systematic drug safety evaluation. - Highlights: • Drugs causing common toxicity lead to similar in vitro gene expression changes. • We built a model to predict drug toxicity with drug-specific expression profiles. • Drugs with FDA black box warnings were effectively identified by our model. • In vitro assay can detect severe toxicity in the early stage of drug development.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2019

Predicting Sooting Tendencies from Chemical Structure with Experimental and Theoretical Insight

Particulate matter (PM) emissions from internal combustion engines negatively impact public health and global climate. These problems are exacerbated by newer gasoline direct injection engines, which are more fuel-efficient, but also produce more soot than traditional spark ignited engines. Reducing soot formation is therefore of paramount importance in the development of new fuels. A fuel's sooting tendency is a quantitative parameter that describes the sooting behavior of a pure compound or fuel mixture. The yield sooting index (YSI), developed by McEnally and Pfefferle, accurately measures sooting tendencies using small sample quantities. Using an experimental sooting tendency database, we have developed a predictive model for sooting behavior from a quantitative structure-activity relationship (QSAR). It was developed so that input molecules are first decomposed into individual carbon-type fragments for which the sooting tendency contribution can be assigned based on a Bayesian linear regression against the experimental database. The model's predictive accuracy is comparable to its training performance using leave-one-out cross-validation. We have used this model to provide quantitative insights into the effects of chemical structure on soot formation, but excitingly, we have also been able to readily identify the presence of more complicated kinetic sooting mechanisms for structures which are extreme outliers. Oxygenated aromatics can be produced readily from biomass as renewable sources and oxygenated aromatics with very similar structures tend to have a much lower sooting tendency, for example methoxybenzene (anisole, 107), 2-methylphenol (m-cresol, 103), 2-ethylphenol (120), 3-ethylphenol (138) and 1-phenylethanol (142). Thus, the presence of just one oxygen atom in an aromatic compound can drastically alter the reaction pathways leading to soot precursors. We have applied density functional theory (DFT) calculations and flow reactor experiments to examine how oxygenation alters reaction pathways in a combustion environment. This study has allowed us to gain understanding on how the location of an oxygenated functional group influences soot formation. Our work provides a blueprint for the design of oxygenated fuels from biomass, which minimize the production of soot in low oxygen environments.

09 BIOMASS FUELS↗
DOE OSTI2016

Using Topic Modeling and Text Embeddings to Predict Deleted Tweets

Predictive models for tweet deletion have been a relatively unexplored area of Twitter-related computational research. We first approach the deletion of tweets as a spam detection problem, applying a small set of handcrafted features to improve upon the current state-of-the- art in predicting deleted tweets. Next, we apply our approach to a dataset of deleted tweets that better reflects the current deletion rate. Since tweets are deleted for reasons beyond just the presence of spam, we apply topic modeling and text embeddings in order to capture the semantic content of tweets that can lead to tweet deletion. Our goal is to create an effective model that has a low-dimensional feature space and is also language-independent. A lean model would be computationally advantageous processing high-volumes of Twitter data, which can reach 9,885 tweets per second. Our results show that a small set of spam-related features combined with word topics and character-level text embeddings provide the best f1 when trained with a random forest model. The highest precision of the deleted tweet class is achieved by a modification of paragraph2vec to capture author identity.

Potash, Peter J.↗
DOE OSTI2019

Characterizing the length-dependence of DNA nanotube end-to-end joining rates

DNA nanotechnology provides a route towards the synthesis of custom nano-structured materials and circuits through hierarchical assembly processes. Although predictive kinetic models are being developed for the assembly of DNA nanostructures from small monomeric components, a general model for the hierarchical assembly of DNA nanostructures remains elusive. DNA tile nanotubes provide an ideal model system for the study of hierarchical assembly via end-to-end joining. In this study, we experimentally characterize the length-dependence of the end-to-end joining rate of DNA tile nanotubes. We then test the ability of three different models of polymer end-to-end joining to reproduce experimentally measured changes in nanotube lengths during a joining reaction using an ODE model for nanotube joining. All three models predict physically realistic joining rates that are consistent with prior measurements, with a length-independent end-to-end joining rate model providing the best fit to the experimental data. A length-independent constant joining rate is consistent with other DNA self-assembly processes across a broad range of length scales and also suggests how tractable models for hierarchical DNA nanostructure could be developed.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2022

Multi-fidelity modeling to predict the rheological properties of a suspension of fibers using neural networks and Gaussian processes

Unveiling the rheological properties of fiber suspensions is of paramount interest to many industrial applications. There are multiple factors, such as fiber aspect ratio and volume fraction, that play a significant role in altering the rheological behavior of suspensions. Three-dimensional (3D) numerical simulations of coupled differential equations of the suspension of fibers are computationally expensive and time-consuming. Machine learning algorithms can be trained on the available data and make predictions for the cases where no numerical data are available. However, some widely used machine learning surrogates, such as neural networks, require a relatively large training dataset to produce accurate predictions. Multi-fidelity models, which combine high-fidelity data from numerical simulations and less expensive lower fidelity data from resources such as simplified constitutive equations, can pave the way for more accurate predictions. Here, we focus on neural networks and the Gaussian processes with two levels of fidelity, i.e., high and low fidelity networks, to predict the steady-state rheological properties, and compare them to the single-fidelity network. High-fidelity data are obtained from direct numerical simulations based on an immersed boundary method to couple the fluid and solid motion. The low-fidelity data are produced by using constitutive equations. Multiple neural networks and the Gaussian process structures are used for the hyperparameter tuning purpose. Results indicate that with the best choice of hyperparameters, both the multi-fidelity Gaussian processes and neural networks are capable of making predictions with a high level of accuracy with neural networks demonstrating marginally better performance.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗