SEARCH · Search NASA

Results for “faster”

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.

SourceResearch connectionFrom yearThrough year

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

At least 253 records · Page 14

All sources NASA NTRS DOE OSTI

DOE OSTI2025

Post-build stress-relief optimization for laser powder bed fusion 316H stainless steel

Nuclear energy remains a critical component of a diversified and efficient energy portfolio, offering reliable, high-capacity, and low-carbon power. However, in the U.S., aging infrastructure and the slow qualification and deployment of advanced materials and manufacturing techniques hinder progress in next-generation reactor technologies. This study explores the application of laser powder bed fusion (LPBF) additive manufacturing for stainless steel 316H, with a focus on optimizing post-build heat treatments to enhance material properties for high-temperature nuclear applications. The research targets the optimization of stress-relief temperatures to alleviate postbuild residual stresses, ensuring improvements in the microstructural corelated properties. A series of microstructural and mechanical evaluations were performed on LPBF-printed SS-316H samples which were subjected to annealing at temperatures varying between 650 °C and 850 °C. X-ray diffraction, scanning electron microscopy, and transmission electron microscopy analyses revealed that increasing the heattreatment temperature accelerated dislocation recovery. Vickers microhardness measurements showed an initial reduction in values, followed by stabilization over extended durations at all the temperatures. While higher temperatures facilitated faster recovery, they also promoted carbide precipitation along grain and solidification cell boundaries, narrowing the safe processing window. In contrast, heat treatment at 650°C preserved the cellular substructure and enabled controlled carbide precipitation over time. In conclusion, these findings highlight the importance of time–temperature optimization and suggest that 650°C for up to 2 h provides the most favorable balance between recovery and carbide control for a stress-relief treatment.

316 stainless steel↗

DOE OSTI2024

Heteroatom anchoring to enhance electrochemical reversibility for high-voltage P2-type oxide cathodes of sodium-ion batteries

P2-type cathode has received extensive attention due to its faster Na+ diffusion and a high theoretical capacity in sodium-ion batteries (SIBs). However, undesirable phase transformations have induced dramatic capacity decay of SIBs during the cycling process. In this study, heteroatom anchoring through Cu/Mg dual doping is introduced into P2-type Na 0.67 Ni 0.33 Mn 0.67 O 2 cathode to enhance high-voltage electrochemical reversibility and modulate interfacial Na + kinetics. Further, the as-prepared Na 0.67 Ni 0.23 Mg 0.05 Cu 0.05 Mn 0.67 O 2 exhibits an outstanding capacity retention (83.4% after 2000 cycles at 10C) and rate performance (73 mAh g -1 at 10C, accounting for 58.7% of that at 0.1 C) over the voltage range of 2.5–4.4 V. Intensive explorations further manifest that the modified mechanism of dual-ion doping strategy is attributed to the synergistic coupling effect of a substantial change in Na occupancy distribution and an increase in oxygen vacancy buffer. Thus, the optimized cathode expedites Na + diffusion and reduces detrimental phase transformation, which favors high-rate performance and long-term cycling stability. This study develops a route to rationally design high-voltage cathode materials for SIBs.

25 ENERGY STORAGE↗

DOE OSTI2025

The temporal onset of associations of cortical proteins with cognitive resilience vary during late life

Background: Cortical proteins associated with cognitive resilience have been identified but their temporal onset in older adults is unknown. We present a multistage approach to first identify cortical proteins associated with cognitive resilience and then examine their associated temporal onset. Methods: We used data from a subset of 1088 decedents from two cohort-studies who had selected reaction monitoring proteomics from the dorsolateral prefrontal cortex, and at least 3 cognitive assessments. Cognition was assessed using a composite derived from 19 tests. We first used linear mixed-effects models to identify cortical proteins associated with cognitive resilience. We then used functional mixed-effects models to examine non-linear associations between proteins and cognitive resilience to identify their temporal onset. Results: Mean age at death was 90 years (SD = 6.4); 69 % were female. On average, cognition started to decline at around 15 years before death, with accelerated decline in the last 7 years. We identified 40 proteins associated with cognitive resilience, of which 17 proteins also showed non-linear associations. Non-linear associations indicated that higher levels of 10 proteins were associated with slower cognitive decline between 23 and 4 years before death. In contrast, higher levels of 7 proteins were associated with faster decline only within the last 7 years before death. Conclusions: Cognitive resilience proteins are differentially related to late-life cognitive aging; the onset of proteins that maintain cognition may begin many years before the onset of proteins that hasten cognitive decline. The temporal onset of cognitive resilience proteins may be crucial for timing efficacious interventions.

Zammit, Andrea↗

DOE OSTI2024

Vascular endothelial growth factor receptor-1 (FLT1) interactions with amyloid-beta in Alzheimer’s disease: A putative biomarker of amyloid-induced vascular damage

We have identified FLT1 as a protein that changes during Alzheimer's disease (AD) whereby higher brain protein levels are associated with more amyloid, more tau, and faster longitudinal cognitive decline. Given FLT1's role in angiogenesis and immune activation, we hypothesized that FLT1 is upregulated in response to amyloid pathology, driving a vascular-immune cascade resulting in neurodegeneration and cognitive decline. We sought to determine (1) if in vivo FLT1 levels (CSF and plasma) associate with biomarkers of AD neuropathology or differ between diagnostic staging in an aged cohort enriched for early disease, and (2) whether FLT1 expression interacts with amyloid on downstream outcomes, such as phosphorylated tau levels and cognitive performance. Additionally, we sought to replicate FLT1 interactions in the brain. The results showed that higher levels of FLT1 in CSF and post-mortem brain tissue related to increased tau, particularly among amyloid positive individuals. These analyses help clarify the potential utility of FLT1 as a biomarker among individuals with evidence of brain amyloidosis.

60 APPLIED LIFE SCIENCES↗

DOE OSTI2025

A machine learning framework for accurate and robust analysis of radiation detector pulses

The microscopic properties of atomic nuclei are used to study various scientific questions. They are essential for understanding the fundamental forces of nature and the chemical evolution of the universe. Detecting decay radiation from radioactive nuclei makes it possible to probe these fundamental nuclear properties. Detector waveform traces may contain additional information about the radiation. Generally, advanced signal processing techniques are needed to extract this additional information, often involving fitting the waveform with model response functions using non-linear least-squares optimization with second-order gradient methods. While this is a powerful technique, it is also computationally expensive, leading to slow processing time, which scales with the volume of data. To address this problem, we have developed a machine learning (ML) approach that infers the characteristics of traces from a model detector response function. In particular, we are interested in classifying whether a single recorded trace consists of one or two pulse constituents and estimating the pulse parameters. Furthermore, our proposed ML method can precisely extract the pulses’ parameters, such as energy and timing information, and accurately classify the pulse multiplicity of a trace. Unlike non-learning-based approaches, our ML approach uses neural networks that are significantly faster at inference, as they do not require any optimization during this stage.

Curve fitting↗

DOE OSTI2026

Robustness of the smartpixels classifier for different simulated sensor geometries and non-ideal detector conditions

Pixel tracking detectors at upcoming collider experiments will see unprecedented charged-particle densities. Real-time data reduction on the detector will enable higher granularity and faster readout, possibly enabling the use of the pixel detector in high-rate online event selection, such as the ATLAS or CMS first-level trigger systems. This data reduction can be accomplished with a neural network (NN) in the readout chip bonded with the sensor that recognizes and rejects tracks with low transverse momentum (p T ) based on the geometrical shape of the charge deposition (“cluster”). To design viable detectors for deployment, the dependence of the NN as a function of the sensor geometry, external magnetic field, irradiation, and noise must be understood. In this paper, we present first studies of the efficiency and data reduction for planar pixel sensors exploring these parameters. For the CMS HL-LHC sensor geometry, we obtain a signal efficiency of (91.9 ± 0.7)% and a data reduction of (29.7 ± 1.0)%. A smaller sensor pitch in the bending direction improves the p T discrimination, but a larger pitch can be partially compensated with detector thickness. Any accumulated radiation damage also changes the cluster shape, reducing the signal efficiency compared to the baseline by approximately 30–60% in absolute terms, but nearly all of the performance can be recovered through retraining of the network and updating the weights. Finally, the impact of noise was investigated, and retraining the network on noise-injected datasets was found to maintain performance within 6% of the baseline network trained and evaluated on noiseless data. •ASIC-compatible track-momentum classifier is robust in realistic detector conditions.•About 90% signal efficiency and 30% data reduction per layer for CMS HL-LHC geometry.•Single-layer signal efficiency increases for smaller pixel pitch or thicker sensors.•Performance with noise or after radiation damage mostly recovered by retraining.

Shekar, Danush [Illinois U., Chicago] (ORCID:00000↗

DOE OSTI2025

A hybrid surrogate modeling framework for the Digital Twin of a Fluoride-salt-cooled High-temperature Reactor (FHR)

While nuclear energy is a non-greenhouse-gas emitting energy source, expensive operational costs due to the high-level of safety requirements decreases their competitiveness in the sustainable energy market. Advanced reactor concepts paired with Digital Twins aim to increase the commercialization gains of nuclear energy by reducing operational costs, increasing reactor reliability and enhancing power generation. To support Digital Twin tasks such as real-time autonomous control, proactive maintenance monitoring or optimizing power demand operations, a fast and accurate virtual representation of the Nuclear Power Plant (NPP) is required. The computational cost of high-fidelity, physics-based models are unsuitable for real-time analysis or scalability. Here, in this work, a hybrid surrogate modeling framework is developed fora Fluoride-salt-cooled High-temperature Reactor (FHR) that leverages physics-inspired models for key reactor components and uses data-driven methods for rapid system state space prediction. The Xenon reactivity feedback model is integrated to inform the surrogate model about the reactor core and the homologous pump theory model is the basis for representing pump degradation. Using a detailed, two dimensional thermal hydraulics model to generate data on the FHR, we train a network of Vectorized Autoregressive Moving-Average with eXogenous input (VARMAX) models to predict the remaining state values. The result is a surrogate model that provides a detailed reactor state representation of 41 system states and a pump degradation analysis. The framework is applied to Load Follows profiles, yielding high accuracy and a speedup that is more than 4000x faster compared to the higher- fidelity thermal hydraulics model, enabling real-time operational intelligence and applications in long horizon predictions. While the surrogate model framework is demonstrated for the particular case of FHR, the hybrid physical/data-driven modeling approach including the network of surrogates and the underlying modularity has the potential to be applied to other physical asset systems.

Digital Twins↗

DOE OSTI2025

Slender-body approach for computing second-order wave loads in the frequency domain

This work presents a slender-body approach to evaluate the second-order wave loads acting on a floating structure in the frequency domain. The approach is in the same spirit as the common use of Morison’s equation to approximate the wave loads without solving the radiation/diffraction problem. To do so, we employ Rainey’s equation, which can be seen as an extension of the inertial part of Morison’s equation to include nonlinear effects. We introduce modifications to Rainey’s formulation in order to evaluate wave kinematics at the mean body position instead of the original approach of considering instantaneous displacements. We also propose a simple approximation to partially account for wave scattering effects on the second-order loads based on the analytical solution of a surface-piercing bottom-mounted vertical circular cylinder. Though limited to structures composed of cylinders, this slender-body approach is orders of magnitude faster than computing second-order wave coefficients with a radiation/diffraction code. We implemented this approach for difference-frequency (slow drift) loads in an open-source frequency-domain floating wind turbine model. We present comparisons against results obtained with radiation/diffraction theory for three reference floating wind turbine designs: the OC3-Hywind spar, the OC4-DeepCwind semisubmersible, and the VolturnUS-S semisubmersible. In general, the results show that the proposed slender-body approach with the correction to approximate wave scattering effects provides useful estimations of the difference-frequency wave loads and the resulting motions of the floater.

17 WIND ENERGY↗

DOE OSTI2024

Impurity-enhanced core valence luminescence via Zn-doping in cesium magnesium chlorides

Scintillators with faster timing capabilities are currently in high demand for use in radiation detection systems in the fields of nuclear and medical physics. The limited number of suitable materials that meet the performance criteria of next generation detection systems presents an opportunity for discovery of new fast scintillator materials. In this work, the effects of doping several ultrafast core-valence luminescent (CVL) scintillators with divalent Zn is explored. Three compounds are investigated – CsMgCl 3 , Cs 2 MgCl 4 , and Cs 3 MgCl 5 – and single crystals of each doped with 5 mol% Zn are grown via the Bridgman method. Additionally, mixing across the full range of concentrations (from 0 % to 100 % Zn) is explored in the Cs 2 Mg 1-x Zn x Cl 4 and Cs 3 Mg 1-x Zn x Cl 5 systems. For low concentrations of Zn, light yields of all three compounds are enhanced (by up to ~60 %) compared to the pure crystals, achieving what we believe to be the brightest known CVL, CsMgCl 3 :Zn 5 % (3400 ± 170 ph/MeV light yield). More importantly, Zn doping does not affect the ultrafast timing properties, with each composition maintaining a single-component decay time around 1–3 ns. A sub-100 ps coincidence time resolution (CTR) is also achieved with CsMgCl 3 :Zn 5 %. The results of this work reveal a new avenue towards obtaining brighter CVL materials, which could open up possibilities for more advanced ultrafast scintillators to be discovered moving forward.

36 MATERIALS SCIENCE↗

DOE OSTI2025

Plasmonic Ag nanocomposite phosphate glasses produced via γ-ray irradiation as reduction route

This paper reports on the impact of γ-ray irradiation (10, 100 kGy) on melt-quenched Ag + -doped phosphate glass and the effects of subsequent thermal processing leading to the production of plasmonic Ag nanocomposites. The γ-irradiated glasses were characterized alongside the pristine by differential scanning calorimetry (DSC), Raman spectroscopy, electron paramagnetic resonance (EPR) spectroscopy, optical absorption, and photoluminescence (PL) spectroscopy. DSC characterization showed consistent glass transition temperatures (T g ) before and after γ-irradiation whereas the crystallization temperatures tended to decrease with increasing γ-ray dose. However, a lack of alteration of the glass network structure was supported by Raman spectroscopy. Room temperature EPR spectra clearly showed the formation of phosphorus oxygen hole center (POHC) defects in the undoped host, in addition to another doublet likely associated with a P 3 defect. The presence of paramagnetic silver species encompassing Ag 2+ and 107/109 Ag 0 atoms was also indicated in the silver-activated glass together with POHC defects. Optical absorption spectra were also consistent with the presence of various radiation-induced centers. Further analyzing the glass absorption edge via Tauc plots suggested the formation of electron center (EC) defects in γ-irradiated samples wherein the silver-doped glass exhibited decreasing band gap energies with increasing γ-ray dose. The PL characterization showed the silver-related radio-PL effect was induced exhibiting broad band emission with two maxima around 500 and 625 nm stemming from various molecular Ag$^{x+}_{n}$ clusters. Emission decay analyses revealed that the longer wavelength emission exhibited slower decay. The highest radiation dose of 100 kGy however resulted in weaker emission and faster decay kinetics attributed to energy transfer between the luminescent silver species and POHC defects. Finally subjecting the γ-irradiated Ag-doped glasses to heat treatment near the T g at 490 °C led to the development of the surface plasmon resonance of Ag nanoparticles (NPs) and the vanishing of the Ag$^{x+}_{n}$ clusters luminescence. In conclusion, the presence of the matrix-related EC defects was deemed accountable for the thermally induced reduction and consequent precipitation of Ag NPs making the plasmonic glasses attractive for photonic applications such as nonlinear optics.

36 MATERIALS SCIENCE↗

DOE OSTI2024

Understanding parton evolution in matter from renormalization group analysis

We perform a renormalization group (RG) analysis of collinear hadron production in deep inelastic scattering on nuclei. We consider the limit where the parent parton energy E is large, while the medium opacity remains small. We identify the fixed order and leading enhanced medium contributions to the semi-inclusive cross sections and derive RG equations that resum multiple emissions near the endpoints of the splitting functions at first order in opacity. These evolution equations treat the same type of radiation enhancement in matter as the modified Dokshitzer-Gribov-Lipatov-Altarelli-Parisi approach, but differ in the way one regulates the collinear divergences. They provide a unique analytic insight into the problem of resummation and a faster and more efficient path to phenomenology. The new RG evolution framework is applied to study fragmentation in eA reactions.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

DOE OSTI2026

Hygrothermal aging effects on polyimide and acrylate-based adhesive materials in high-performance flat-flex cable assemblies

Flat-flex assemblies are widely used across a diverse range of technologies, including consumer electronics, automotive systems, aerospace and defense applications, and medical devices, due to their compact form factor and flexibility. However, environmental stressors like temperature and humidity are known to degrade performance over time, though the main mechanisms for degradation remain unknown. Here, in this work, we examined the aging behavior of copper/polyimide/adhesive laminate systems under varying temperature and humidity conditions, focusing on material degradation and its effects on mechanical and dielectric properties. Results show that peel strength and breakdown voltage decrease with exposure time, temperature, and humidity; spectroscopic characterization revealed that the adhesive, and not the polyimide, is the weak link in the material stack-up. The adhesive, identified as a butyl acrylate-acrylonitrile (BA-AN) copolymer, exhibited age-related spectral changes that correlated with exposure severity and duration. Hydrolysis at BA ester and AN nitrile groups was identified as the primary degradation mechanism, producing amides, acids, and alcohols. We developed a mechanistic model that connects kinetic parameters for BA-AN spectral band decay to peel strength and breakdown voltage, and implicates hydrolysis at the AN moiety, not BA, in performance decrements. Modeling predictions at ambient conditions indicate faster decline in peel strength than dielectric strength at early times, followed by plateauing of both properties at longer times. These findings provide critical insights into laminate aging mechanisms and highlight the importance of addressing BA-AN degradation to improve the long-term reliability of these systems in high-humidity environments.

organic↗

DOE OSTI2024

Modeling hyperbranched polymer formation via ATRP using dissipative particle dynamics

Hyperbranched polymers (HBPs) offer distinguishing, advantageous properties that arise from their distinctive complex topology. One of the effective approaches to the synthesis of hyperbranched structures involves the use of a branching initiator (inibramer) that is activated only after incorporation into a polymer chain. There remain, however, challenges in determining and characterizing the structures of the synthesized HBPs. Dissipative particle dynamics (DPD) was used to probe the effects of inibramer concentration, solvent concentration, and inibramer reactivity on the kinetics, molecular weight, and dispersity of HBPs. Additionally, DPD allows for direct observation of branched structures, which was not possible in previously reported Monte Carlo type simulations. It was found that higher inibramer concentrations led to faster monomer consumption while forming more dendritic structures with fewer defects. Additionally, high dispersities characteristic of HBP systems were found to originate from asymmetric propagation rates between inibramer-inibramer and inibramer-monomer reactions.

Biswas, Santidan↗

DOE OSTI2024

Understanding the ignition process and flame structure of conventional and oxygenated fuels under engine relevant conditions – An optical study

Renewably generated synthetic fuels such as poly-oxymethylene ethers (OME) have a significant potential to effectively break the soot-NOX trade-off in compression ignition engines by using exhaust gas recirculation (EGR) to maintain low nitrogen oxide (NOX) emissions while maintaining good efficiency and simultaneously contributing to circular carbon economy. However, owing to the fundamental differences in properties of OME when compared to fossil-based diesel fuels, it is critical to fully understand its ignition and combustion phenomenology to take advantage of this fuel to its utmost potential. In this context, this work outlines the results of a systematic experimental study performed in a heavy-duty, single-cylinder, optical engine probing the spatial and temporal progression of fuel decomposition and ignition behavior of OME when compared to n-dodecane, a diesel-fuel surrogate. Thermodynamic analysis and optical diagnostics techniques including simultaneous HCHO-PLIF and OH-PLIF complemented by high-speed OH* chemiluminescence were employed along with parametric sweeps of intake temperature and EGR dilution rates. OME does not exhibit any observable low temperature heat release irrespective of the ambient oxygen concentration. Differences in the observed diffusive flame structure such as longer flame lift-off length, less pronounced combustion recession, faster premixed burn at ignition (“volumetric” ignition), non-sooting behavior suggest that the inherent presence of fuel-bound oxygen in OME can skew the air-fuel ratio (AFR) distribution within the jet thereby reducing the reliance of combustion on mixing and air entrainment. This leads to rapid late-cycle oxidation leading to shorter combustion duration and favorable combustion phasing. Results also suggest that OME exhibits relatively weak negative temperature coefficient (NTC) behavior, however, the OME fuel-decomposition kinetic-pathways produce significant concentration of HCHO, which might be erroneously interpreted as a product of cool-flames.

Air-fuel ratio↗

DOE OSTI2025

Application of soot carbonization kinetics to deduce meaningful soot formation rates in premixed flat flames

Soot formation rates measured in fuel-rich premixed flat flames are frequently used to calibrate or validate chemical kinetic models of soot formation. Unfortunately, these flames feature an extended region of soot precursor particle inception and carbonization that complicates interpretation of soot measurements and leads to a fundamental inconsistency in the nature of the soot material that is modeled versus what is being measured when using non-intrusive, optical techniques. In the work presented here, previously reported data on two canonical sooting ethylene-air premixed flames at 1 atm pressure are interpreted via a new analysis approach that combines soot optical dispersion coefficient measurements with soot carbonization kinetics. This analytical approach explicitly accounts for the production of poorly ordered soot precursor particle mass and its carbonization over time in the flames, providing a clear distinction between the formation rate of precursor particles and their transformation into ordered, solid soot particulate mass. In particular, the results of the analysis show that the precursor particles form much earlier in the flame than the majority of the carbonized soot and their formation rate is two to three times faster than that of ordered soot. The results also show that particle agglomeration begins when the particles are at an intermediate state of carbonization. In conclusion, these results offer a valuable new interpretation of these important datasets and should lead to substantial improvements in the development and calibration of quantitative soot models.

Soot formation↗

DOE OSTI2025

Influence of americium complexation on the radiation-induced chemical reactivity of sulfophenyl bistriazinyl pyridine (SO 3 -Ph-BTP) towards the nitrate radical

Sulfophenyl bistriazinyl pyridine (SO 3 -Ph-BTP) is a hydrophilic organic ligand used to separate actinides and lanthanides. Dose accumulation and time-resolved studies have previously provided insight into the radiolytic stability and degradation pathways of SO 3 -Ph-BTP in reprocessing environments, but no study has yet addressed the impact of minor actinide complexation on the radiation chemistry of this ligand. To begin to fill this knowledge gap, a systematic, time-resolved study exploring the reactivity of the nitrate radical (NO 3 • ) with SO 3 -Ph-BTP in the presence of trivalent americium, Am(III), has been conducted, which demonstrates enhanced reactivity (an order of magnitude faster) upon metal complexation.

38 - RADIATION CHEMISTRY, RADIOCHEMISTRY, AND NUCL↗

DOE OSTI2024

Near wake evolution of a tidal stream turbine due to asymmetric sheared turbulent inflow with different integral length scales

Tidal stream turbines deployed at highly energetic open water sites are subjected to sheared inflow in the rotor plane. The inflow shear is expected to cause asymmetric loading on the rotor blades and affect the downstream wake. In the current study, two different turbulent inflow conditions, static-high shear and dynamic shear, were generated via an active-grid turbulence generator. A 1:20 scaled three-bladed horizontal axis tidal turbine model was tested in those conditions. The results were compared to a quasi-laminar case with no imposed turbulence or shear. The results show that the high shear reduces the average performance, with a drop of up to 16% in the optimal power coefficient. Besides, the shear profiles increase torque fluctuations and induce significant differences in wake hydrodynamics between the high-speed (upper) and low-speed (lower) regions. The large integral length scales further enhance the load fluctuations perceived by the rotor but have a negligible effect on the mean wake field quantities and the wake recovery. The lower half region featured a faster breakdown of tip vortex structure and a rapid drop of swirl number, a phenomenon conjectured to be a consequence of the strong turbulence intensities and Reynolds stresses in the lower half region. Furthermore, the sheared turbulent inflow also results in a very intensive energy redistribution process towards large-scale, low-frequency motions, which is important to the downstream turbines.

16 TIDAL AND WAVE POWER↗

DOE OSTI2024

Fourier-MIONet: Fourier-enhanced multiple-input neural operators for multiphase modeling of geological carbon sequestration

Geologic carbon sequestration (GCS) is a safety-critical technology that aims to reduce the amount of carbon dioxide in the atmosphere, which also places high demands on reliability. Multiphase flow in porous media is essential to understand CO 2 migration and pressure fields in the subsurface associated with GCS. However, numerical simulation for such problems in 4D is computationally challenging and expensive, due to the multiphysics and multiscale nature of the highly nonlinear governing partial differential equations (PDEs). It prevents us from considering multiple subsurface scenarios and conducting real-time optimization. Here, we develop a Fourier-enhanced multiple-input neural operator (Fourier-MIONet) to learn the solution operator of the problem of multiphase flow in porous media. Fourier-MIONet utilizes the recently developed framework of the multiple-input deep neural operators (MIONet) and incorporates the Fourier neural operator (FNO) in the network architecture. Once Fourier-MIONet is trained, it can predict the evolution of saturation and pressure of the multiphase flow under various reservoir conditions, such as permeability and porosity heterogeneity, anisotropy, injection configurations, and multiphase flow properties. Compared to the enhanced FNO (U-FNO), the proposed Fourier-MIONet has 90% fewer unknown parameters, and it can be trained in significantly less time (about 3.5 times faster) with much lower CPU memory (<15%) and GPU memory (<35%) requirements, to achieve similar prediction accuracy. In addition to the lower computational cost, Fourier-MIONet can be trained with only 6 snapshots of time to predict the PDE solutions for 30 years. Furthermore, we observed that Fourier-MIONet can maintain good accuracy when predicting out-of-distribution (OOD) data. The excellent generalizability of Fourier-MIONet is enabled by its adherence to the physical principle that the solution to a PDE is continuous over time. Furthermore, the developed Fourier-MIONet makes it possible to solve the long-time evolution of geological carbon sequestration in a large-scale three-dimensional space accurately and efficiently.

97 MATHEMATICS AND COMPUTING↗