Search NASA⌕ Search

SEARCH · Search NASA

Results for “consistent scheme”

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 595 records · Page 33

Limitations of Bin and Bulk Microphysics in Reproducing the Observed Spatial Structure of Light Precipitation

Abstract Coarse-gridded atmospheric models often account for subgrid-scale variability by specifying probability distribution functions (PDFs) of process rate inputs such as cloud and rainwater mixing ratios ( q c and q r , respectively). PDF parameters can be obtained from numerous sources: in situ observations, ground- or space-based remote sensing, or fine-scale modeling such as large-eddy simulation (LES). LES is appealing to constrain PDFs because it generates large sample sizes, can simulate a variety of cloud regimes/case studies, and is not subject to the ambiguities of observations. However, despite the appeal of using model output for parameterization development, it has not been demonstrated that LES satisfactorily reproduces the observed spatial structure of microphysical fields. In this study, the structure of observed and modeled microphysical fields are compared by applying bifractal analysis, an approach that quantifies variability across spatial scales, to simulations of a drizzling stratocumulus field that span a range of domain sizes, drop concentrations (a proxy for mesoscale organization), and microphysics schemes (bulk and bin). Simulated q c closely matches observed estimates of bifractal parameters that measure smoothness and intermittency. There are major discrepancies between observed and simulated q r properties, though, with bulk simulated q r consistently displaying the bifractal properties of observed clouds (smooth, minimally intermittent) rather than rain while bin simulations produce q r that is appropriately intermittent but too smooth. These results suggest fundamental limitations of bulk and bin schemes to realistically represent higher-order statistics of the observed rain structure.

54 ENVIRONMENTAL SCIENCES↗

Probing Complexity using the Linac Coherent Light Source (LCLS) and the Advanced Light Source (ALS)

The goal of our research program is to investigate fundamental interactions between photons and molecular/nano-systems to advance our quantitative understanding of electron correlations, charge transfer and many body phenomena. Our research projects focus on probing, on a femtosecond time-scale, multi-electron interactions and tracing nuclear motion in order to understand, and ultimately control energy flow and charge transfer processes from electromagnetic radiation to matter. The experiments will be carried out with state of the art instrumentation built by the P.I. team with funds from a DoE "Single Investigator and Small Group Research" (SISGR) grant. The research projects carried out the past three years consisted of first experiments using the Linac Coherent Light Source (LCLS) x-ray free electron laser (FEL) facility at the SLAC National Laboratory, as well as the study of correlated processes in select anions using the Advanced Light Source (ALS). A report for the past cycle is described in section II. These studies have paved the way for our renewal application for the next three years. Our research interests for the next three years extend our past and present research by carrying out time-resolved measurements described in section III. They will consist of: a) The study of molecular dynamics that happen on ultrafast time scales, using pump-probe schemes and the study of non-linear physics in the x-ray regime via multi-photon absorption from the LCLS. This will be achieved by measuring and examining both electronic and nuclear dynamics subsequent to the interaction of molecules and nano-systems with LCLS pulses of various wavelength, intensity and pulse duration as described in section III.A. b) The study of molecular dynamics and correlated processes via absorption of vuv-soft x-rays from the Advanced Light Source (ALS) at Lawrence Berkeley Laboratory to provide single-photon ionization baseline results for LCLS studies. In addition, we will study the photodetachment of anions using the ALS as described in section III. B. The PI has an ALS approved program (AP) thus has guaranteed beamtime. Our research underpins several aspects of the DOE mission since our investigations of the formation and evolution of energized states in molecules and nanostructures provides a fundamental basis for understanding elementary processes in solar energy conversion and radiation-induced chemistry. Most of our work is carried out in a strong partnership with theorists. Our proposed work falls in the class of research reported to be timely and necessary by the 2013 DoE Basic Energy Sciences Advisory Committee (BESAC) Subcommittee report on Future X-ray Light Sources. Specifically, our proposed experiments fall in the category labeled class II, “Spectroscopy and Dynamics with FELs”.

74 ATOMIC AND MOLECULAR PHYSICS↗

Shadow molecular dynamics and atomic cluster expansions for flexible charge models

A shadow molecular dynamics scheme for flexible charge models is presented, where the shadow Born-Oppenheimer potential is derived from a coarse-grained approximation of range-separated density functional theory. The interatomic potential, including the atomic electronegativities and the charge-independent short-range part of the potential and force terms, are modeled by the linear atomic cluster expansion (ACE), which provides a computationally efficient alternative to many machine learning methods. The shadow molecular dynamics scheme is based on extended Lagrangian (XL) Born-Oppenheimer molecular dynamics (BOMD) [Eur. Phys. J. B 94, 164 (2021)]. XL-BOMD provides a stable dynamics, while avoiding the costly computational overhead associated with solving an all-to-all system of equations, which normally is required to determine the relaxed electronic ground state prior to each force evaluation. To demonstrate the proposed shadow molecular dynamics scheme for flexible charge models using the atomic cluster expansion, we emulate the dynamics generated from self-consistent charge density functional tight-binding (SCC-DFTB) theory using a second-order charge equilibration (QEq) model. The charge-independent potentials and electronegativities of the QEq model are trained for a supercell of uranium oxide (UO2) and a molecular system of liquid water. The combined ACE + XL-QEq dynamics are stable over a wide range of temperatures both for the oxide and the molecular systems, and provide a precise sampling of the Born-Oppenheimer potential energy surfaces. Accurate ground Coulomb energies are produced by the ACE-based electronegativity model during an NVE simulation of UO 2 , predicted to be within 1 meV of those from SCC-DFTB on average during comparable simulations.

74 ATOMIC AND MOLECULAR PHYSICS↗

Experimental Study for Validation of Resilient Response of Controls - Thrust 2 Controls Assessment

The focus of this project is to design, execute, and analyze experiments around resilience response. As part of the Resilience through Data-Driven, Intelligently Designed Control (RD2C) initiative, this project builds upon the previously designed control algorithms and experiments. It designs detailed uses cases to test RD2C-developed mitigation schemes in specific realistic scenarios, analyze and save the measurement data sets. Having a high-confidence and consistent data set is crucial for identifying relevant metrics to describe the system resilience in specific real-life-like scenarios of power grid operation.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Structure of Odd-A Ag Isotopes Studied via Algebraic Approaches

The structure of the odd-A silver isotopes Ag 103–115 is discussed within the frame of the interacting boson–fermion model (IBFM). An overview of their key properties is presented, with a particular attention paid to the “ J -1 anomaly”, represented by an abnormal ordering of the lowest 7/2 + and 9/2 + states. By examining previously published data and newly performed calculations, it is demonstrated that the experimentally known level schemes and electromagnetic properties of Ag 103–115 can be reproduced well within IBFM-1 by using a consistent set of model parameters. The contribution of different single-particle orbitals to the structure of the lowest-lying excited nuclear states in Ag 103–115 is discussed. Given that the J-1 anomaly brings down the 7/2 + level from the j −3 multiplet to energies, which can be thermally populated in hot stellar environments, the importance of low-lying excited states in odd-A silver isotopes for astrophysical processes is outlined.

IBFM-1↗

Improving Bulk Microphysics Parameterizations in Simulations of Aerosol Effects

To improve the microphysical parameterizations for simulations of the aerosol indirect effect (AIE) in regional and global climate models, a double-moment bulk microphysical scheme presently implemented in the Weather Research and Forecasting (WRF) model is modified and the results are compared against atmospheric observations and simulations produced by a spectral bin microphysical scheme (SBM). Rather than using prescribed aerosols as in the original bulk scheme (Bulk-OR), a prognostic doublemoment aerosol representation is introduced to predict both the aerosol number concentration and mass mixing ratio (Bulk-2M). The impacts of the parameterizations of diffusional growth and autoconversion and the selection of the embryonic raindrop radius on the performance of the bulk microphysical scheme are also evaluated. Sensitivity modeling experiments are performed for two distinct cloud regimes, maritime warm stratocumulus clouds (SC) over southeast Pacific Ocean from the VOCALS project and continental deep convective clouds (DCC) in the southeast of China from the Department of Energy/ARM Mobile Facility (DOE/AMF) - China field campaign. The results from Bulk-2M exhibit a much better agreement in the cloud number concentration and effective droplet radius in both the SC and DCC cases with those from SBM and field measurements than those from Bulk-OR. In the SC case particularly, Bulk-2M reproduces the observed drizzle precipitation, which is largely inhibited in Bulk-OR. Bulk-2M predicts enhanced precipitation and invigorated convection with increased aerosol loading in the DCC case, consistent with the SBM simulation, while Bulk-OR predicts the opposite behaviors. Sensitivity experiments using four different types of autoconversion schemes reveal that the autoconversion parameterization is crucial in determining the raindrop number, mass concentration, and drizzle formation for warm 2 stratocumulus clouds. An embryonic raindrop size of 40 μm is determined as a more realistic setting in the autoconversion parameterization. The saturation adjustment employed in calculating condensation/evaporation in the bulk scheme is identified as the main factor responsible for the large discrepancies in predicting cloud water in the SC case, suggesting that an explicit calculation of diffusion growth with predicted supersaturation is necessary for further improvements of the bulk microphysics scheme. Lastly, a larger rain evaporation rate below cloud is found in the bulk scheme in comparison to the SBM simulation, which could contribute to a lower surface precipitation in the bulk scheme.

aerosol↗

Texture simulation of a severely cold rolled low carbon steel using polycrystal modeling

Characteristic textures of cold-rolled low carbon steel consist of the RD//$\langle$110$\rangle$ fiber (α fiber) and ND//$\langle$111$\rangle$ fiber (γ fiber). Although the development of the γ fiber has been successfully modeled by the relaxed constraints polycrystal plasticity model, modeling of the α fiber has been less successful. Here, we investigate mechanisms underlying the development of a strong α fiber in a low carbon steel during cold rolling to large thickness reductions of 90%, 98%, and 99.8%. We compare the deformation textures measured using X-ray and electron backscatter diffraction methods with those simulated using polycrystal models, employing one of four homogenization schemes: the Taylor model, Secant model, Tangent model and an intermediate stiffness model between the Secant and Tangent models. The measured textures consist of an extremely strong α fiber and unusually weak γ fiber. The calculated textures achieve the best agreement with the measured ones when (1) the intermediate stiffness model is used to account for grain shape effects, (2) the flow rule exponent (n) is chosen to allow for multiple slip in grains, (3) latent hardening effects are suppressed, (4) $\langle$111$\rangle$ slip on {110}, {112} and {123} planes is made available for deformation, and (5) the grains shape is permitted to evolve individually. Taken together, these aspects represent indirectly the effects of cross slip on slip activity, suggesting that the strong α fiber in severely cold rolled steel results from cross slip. Although the cross slip is a well accepted mechanism in b.c.c. crystals, its connection to α fiber dominance in texture development during severe rolling deformation has not been established before.

36 MATERIALS SCIENCE↗

Quantum dot polarisation converter in an optomechanical cavity

We propose a scheme of a quantum photon polarisation converter, which is based on controlled electron – photon – phonon transitions in a hybrid semiconductor nanostructure. This structure consists of a GaAs/InAs quantum dot (QD) that has a parallelepiped shape and contains a single electron, and an optomechanical microcavity (MC) based on a photonic crystal (PC) that supports two orthogonally polarised photonic modes and one mechanical (phonon) mode. Within the framework of the microscopic theory, the QD and MC performance characteristics are found. Populations of states of the system as functions of time and its parameters are calculated. The principal possibility of photon polarisation conversion using transitions in a five-level resonance scheme for coherent (single-photon) and steady-state (subphoton) regimes is shown. The MC optical and mechanical spectra are simulated, and the PC structure parameters are selected to ensure the efficient operation of the converter. (paper)

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

On the design of stable, consistent, and conservative high-order methods for multi-material hydrodynamics

Obtaining stable and high-order numerical solutions for multi-material hydrodynamics is an open challenge. Although slope limiters are widely used to maintain monotonicity near discontinuities, typical limiting procedures violate closure laws at the discrete level when applied to multi-material hydrodynamics equations. Due to this, the high-order expansions of quantities related by the closure laws are no longer consistent. The commonly observed symptom of this consistency-violation is that the numerical method fails to maintain constant pressure and velocity across material interfaces. This leads to sub-optimal convergence rates for smooth multi-material problems as well. Specialized limiting procedures that satisfy consistency while maintaining conservation need to be developed for such equations. A novel procedure that re-instates consistency into slope-limited high-order discretizations applied to the multi-material hydrodynamics equations is presented here. Using simple examples, it is demonstrated that the presented method satisfies closure laws at the discrete level, while maintaining conservative properties of the high-order method. Furthermore, this procedure involves a projection step which relies on the compact basis of the underlying spatial discretization, i.e. for discontinuous schemes (viz. DG and FV) the projection is local, and does not involve global matrix solves. Comparisons with conventional approaches emphasizes the necessity of the consistent closure-law preserving limiting approach, in order to maintain design order of accuracy for smooth multi-material problems.

36 MATERIALS SCIENCE↗

A new ionospheric electron precipitation module coupled with RAM-SCB within the geospace general circulation model

Electron precipitation down to the atmosphere due to wave-particle scattering in the magnetosphere contributes significantly to the auroral ionospheric conductivity. In order to obtain the auroral conductivity in global MHD models that are incapable of capturing kinetic physics in the magnetosphere, MHD parameters are often used to estimate electron precipitation flux for the conductivity calculation. Such an MHD approach, however, lacks self-consistency in representing the magnetosphere-ionosphere coupling processes. In this study we improve the coupling processes in global models with a more physical method. We calculate the physics-based electron precipitation from the ring current and map it to the ionospheric altitude for solving the ionospheric electrodynamics. In particular, we use the BATS-R-US (Block Adaptive Tree Scheme-Roe type-Upstream) MHD model coupled with the kinetic ring current model RAM-SCB (Ring current-Atmosphere interaction Model with Self-Consistent Magnetic field (B)) that solves pitch angle-dependent electron distribution functions, to study the global circulation dynamics during the 25–26 January 2013 storm event. Since the electron precipitation loss is mostly governed by wave-particle resonant scattering in the magnetosphere, we further investigate two loss methods of specifying electron precipitation loss associated with wave-particle interactions: (1) using pitch angle diffusion coefficients D αα (E,α) determined from the quasi-linear theory, with wave spectral and plasma density obtained from statistical observations (named as “diffusion coefficient method”) and (2) using electron lifetimes τ(E) independent on pitch angles inferred from the above diffusion coefficients (named as “lifetime method”). We found that both loss methods demonstrate similar temporal evolution of the trapped ring current electrons, indicating that the impact of using different kinds of loss rates is small on the trapped electron population. Furthermore, for the precipitated electrons, the lifetime method hardly captures any precipitation in the large L shell (i.e., 4 < L < 6.5) region, while the diffusion coefficient method produces much better agreement with NOAA/POES measurements, including the spatial distribution and temporal evolution of electron precipitation in the region from the premidnight through the dawn to the dayside. Further comparisons of the precipitation energy flux to DMSP observations indicates that the new physics-based precipitation approach using diffusion coefficients for the ring current electron loss can explain the diffuse electron precipitation in the dawn sector, such as the enhanced precipitation flux at auroral latitudes and flux drop near the subauroral latitudes, but the traditional MHD approach largely overestimates the precipitation flux at lower latitudes.

58 GEOSCIENCES↗

The Principle of Maximum Conformality

A key problem in making precise perturbative QCD predictions is the uncertainty in determining the renormalization scale of the running coupling {alpha}{sub s}({mu}{sup 2}). It is common practice to guess a physical scale {mu} = Q which is of order of a typical momentum transfer Q in the process, and then vary the scale over a range Q/2 and 2Q. This procedure is clearly problematic since the resulting fixed-order pQCD prediction will depend on the renormalization scheme, and it can even predict negative QCD cross sections at next-to-leading-order. Other heuristic methods to set the renormalization scale, such as the 'principle of minimal sensitivity', give unphysical results for jet physics, sum physics into the running coupling not associated with renormalization, and violate the transitivity property of the renormalization group. Such scale-setting methods also give incorrect results when applied to Abelian QED. Note that the factorization scale in QCD is introduced to match nonperturbative and perturbative aspects of the parton distributions in hadrons; it is present even in conformal theory and thus is a completely separate issue from renormalization scale setting. The PMC provides a consistent method for determining the renormalization scale in pQCD. The PMC scale-fixed prediction is independent of the choice of renormalization scheme, a key requirement of renormalization group invariance. The results avoid renormalon resummation and agree with QED scale-setting in the Abelian limit. The PMC global scale can be derived efficiently at NLO from basic properties of the PQCD cross section. The elimination of the renormalization scheme ambiguity using the PMC will not only increases the precision of QCD tests, but it will also increase the sensitivity of colliders to new physics beyond the Standard Model.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Self-consistent field theory based molecular dynamics with linear system-size scaling

We present an improved field-theoretic approach to the grand-canonical potential suitable for linear scaling molecular dynamics simulations using forces from self-consistent electronic structure calculations. It is based on an exact decomposition of the grand canonical potential for independent fermions and does neither rely on the ability to localize the orbitals nor that the Hamilton operator is well-conditioned. Hence, this scheme enables highly accurate all-electron linear scaling calculations even for metallic systems. The inherent energy drift of Born-Oppenheimer molecular dynamics simulations, arising from an incomplete convergence of the self-consistent field cycle, is circumvented by means of a properly modified Langevin equation. The predictive power of the present approach is illustrated using the example of liquid methane under extreme conditions.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

B - L violating nucleon decay and GUT scale baryogenesis in SO(10)

Minimal SO(10) models which utilize a 126-plet of Higgs for breaking B - L symmetry are predictive schemes for neutrino masses and mixings. These models have been known to predict sin{sup 2}2{theta}{sub 13} Asymptotically-Equal-To 0.09, which is consistent with recent measurements. It is shown that the same class of models generates naturally baryon number violating dimension seven operators leading to novel nucleon decay modes such as n{yields}e{sup -}K{sup +}, e{sup -}{pi}{sup +} and p{yields}{nu}{pi}{sup +} which violate B - L by two units. We find that in two-step breaking schemes of non-supersymmetric SO(10), the partial lifetimes for these modes can be within reach of experiments. The interactions responsible for these decay modes also provide a new way to understand the origin of matter in the universe via the decays of GUT scale colored scalar bosons. Their (B - L)-violating nature guarantees that the GUT scale induced baryon asymmetry is not washed out by the electroweak sphaleron interactions. In minimal SO(10) models this asymmetry is closely tied to the masses of quarks, leptons and the neutrinos.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

A Hierarchical Multi-Agent Based Anomaly Detection for Wide-Area Protection in Smart Grid

Future smart grid capabilities provide assurance to expand the advanced information and communication technologies to evolve into densely interconnected cyber physical system. Remedial Action Scheme (RAS), widely used for wide-area protection, relies on the interconnected networks and data sharing devices, which are exposed to the multitude of vulnerabilities. This paper presents our proposed approach to developing multi-agent based RAS scheme against the system-aware stealthy cyber-attacks. Specifically, we propose the two-level hierarchical architecture which consists of distributed local RAS controllers (RAScs) as local agents, operating at different zones/ areas, which are constantly monitored by an overseer, the central agent. The local controllers receive local and randomly changing outside zonal measurements and cyclically forwards to the overseer. The overseer identifies the corrupted controller using the anomaly detection algorithm which processes the measurements coming from the local controllers, compute measurement errors using local and outside zonal measurements, perform validation checks, and finally detect anomalies based on the two-step verification. Next, as a proof of concept, we have implemented and validated the proposed methodology in cyber physical environment at Iowa State’s PowerCyber testbed. We have also implemented the coordinated attack vectors which involve corrupting the local controller and later performing stealthy attacks on the system’s generator. We have evaluated its performance during the online testing in terms of detection rate and Iatency. The experimental results show that it is efficient in detecting different classes of attacks, including ramp and pulse attacks.

24 POWER TRANSMISSION AND DISTRIBUTION↗

A numerical Poisson solver with improved radial solutions for a self-consistent locally scaled self-interaction correction method

Abstract The universal applicability of density functional approximations is limited by self-interaction error made by these functionals. Recently, a novel one-electron self-interaction-correction (SIC) method that uses an iso-orbital indicator to apply the SIC at each point in space by scaling the exchange-correlation and Coulomb energy densities was proposed. The locally scaled SIC (LSIC) method is exact for the one-electron densities, and unlike the well-known Perdew–Zunger SIC (PZSIC) method recovers the uniform electron gas limit of the uncorrected density functional approximation, and reduces to PZSIC method as a special case when isoorbital indicator is set to the unity. Here, we present a numerical scheme that we have adopted to evaluate the Coulomb potential of the electron density scaled by the iso-orbital indicator required for the self-consistent LSIC calculations. After analyzing the behavior of the finite difference method (FDM) and the green function solution to the radial part of the Poisson equation, we adopt a hybrid approach that uses the FDM for the Coulomb potential due to the monopole and the GF for all higher-order terms. The performance of the resultant hybrid method is assessed using a variety of systems. The results show improved accuracy than earlier numerical schemes. We also find that, even with a generic set of radial grid parameters, accurate energy differences can be obtained using a numerical Coulomb solver in standard density functional studies.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

The CSIRO Mk3L climate system model v1.0 coupled to the CABLE land surface scheme v1.4b: evaluation of the control climatology

The CSIRO Mk3L climate system model, a reduced-resolution coupled general circulation model, has previously been described in this journal. The model is configured for millennium scale or multiple century scale simulations. This paper reports the impact of replacing the relatively simple land surface scheme that is the default parameterisation in Mk3L with a sophisticated land surface model that simulates the terrestrial energy, water and carbon balance in a physically and biologically consistent way. An evaluation of the new model s near-surface climatology highlights strengths and weaknesses, but overall the atmospheric variables, including the near-surface air temperature and precipitation, are simulated well. The impact of the more sophisticated land surface model on existing variables is relatively small, but generally positive. More significantly, the new land surface scheme allows an examination of surface carbon-related quantities including net primary productivity which adds significantly to the capacity of Mk3L. Overall, results demonstrate that this reduced-resolution climate model is a good foundation for exploring long time scale phenomena. The addition of the more sophisticated land surface model enables an exploration of important Earth System questions including land cover change and abrupt changes in terrestrial carbon storage.

54 ENVIRONMENTAL SCIENCES↗

Scale-fixed predictions for γ + η c production in electron-positron collisions at NNLO in perturbative QCD

In the paper, we present QCD predictions for γ + η c production at an electron-positron collider up to next-to-next-to-leading order (NNLO) accuracy without renormalization scale ambiguities. The NNLO total cross-section for e + + e – → γ + η c using the conventional scale-setting approach has large renormalization scale ambiguities, usually estimated by choosing the renormalization scale to be the e + e – center-of-mass collision energy $\sqrt{s}$. The Principle of Maximum Conformality (PMC) provides a systematic way to eliminate such renormalization scale ambiguities by summing the nonconformal β contributions into the QCD coupling α s ( Q 2 ). The renormalization group equation then sets the value of α s for the process. The PMC renormalization scale reflects the virtuality of the underlying process, and the resulting predictions satisfy all of the requirements of renormalization group invariance, including renormalization scheme invariance. After applying the PMC, we obtain a renormalization scale-and-scheme independent prediction, σ | NNLO,PMC ≃ 41.18 fb for $\sqrt{s}$=10.6 GeV. The resulting pQCD series matches the series for conformal theory and thus has no divergent renormalon contributions. The large K factor which contributes to this process reinforces the importance of uncalculated NNNLO and higher-order terms. Using the PMC scale-and-scheme independent conformal series and the Padé approximation approach, we predict σ | NNNLO,PMC+Pade ≃ 18.99 fb, which is consistent with the recent BELLE measurement ${\sigma}^{\mathrm{obs}}={16.58}_{-9.93}^{+10.51}$ fb at $\sqrt{s}$ ≃ 10.6 GeV. This procedure also provides a first estimate of the NNNLO contribution.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

GW-PINN: A deep learning algorithm for solving groundwater flow equations

Machine learning methods provide new perspective for more convenient and efficient prediction of groundwater flow. In this study, a deep learning method “GW-PINN” without labeled data for solving groundwater flow equations with wells was proposed. GW-PINN takes the physics inform neural network (PINN) as the backbone and uses either the hard or soft constraint in the loss function for training. A locally refined sampling strategy (LRS) is adopted to generate the consistent spatial sampling points for problems with strong hydraulic head change, and then combined with an appropriate temporal sampling scheme to obtain the final spatial-temporal sampling points. A snowball-style two-stage training strategy by dividing the temporal domain into two subdomains is designed to decrease the sampling points. Five cases were designed to test the training performance of GW-PINN under different sampling strategies and two constraints. The predicted results of GW-PINN were compared with MODFLOW and the analytical solution. The results demonstrate that GW-PINN possesses strong ability in capturing the hydraulic head change for both confined and un-confined aquifers. The hard constraint owns more robust learning ability than the soft constraint. The LRS strategy can generate more accurate results with much fewer sampling points than traditional sampling strategies, and the snowball-style two-stage training strategy is significantly efficient for problems with the drastic change of hydraulic head. Additionally, the application of GW-PINN as a surrogate model for parameterized groundwater flow equations is illustrated. This study provides an option tool for efficient groundwater flow simulation, especially for those with local refinements are needed.

54 ENVIRONMENTAL SCIENCES↗