Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Numerical”

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 631 records · Page 35
All sourcesNASA NTRSDOE OSTI
DOE OSTI2019

Experimental and numerical evaluations of the energy savings potential of thermally anisotropic composites

Previous numerical studies conducted in FY 17 showed that the application of thermally anisotropic composites (TACs) coupled with a heat sink/source can re-direct and reduce unwanted heat flows through exterior building walls, with associated reductions in envelope-generated heating and cooling loads in buildings. The numerical analyses included two-dimensional finite element models (FEM) of wall assemblies and whole-building modeling to estimate the annual reductions in wall-generated heating and cooling loads, and associated energy consumption. The FY17 study used combined FEM and EnergyPlus (E+) simulations of a typical two-story, single-family DOE prototype residential building model.1 We studied performance of several configurations of walls containing exterior continuous insulation and thermally anisotropic composites on the wall exterior. The study showed that annual energy cost savings of up to $270 are possible under Phoenix and Baltimore weather conditions, by retrofitting a baseline wall with an anisotropic composite. In the cases considered, the TAC consisted of three layers of aluminum foil-faced foam boards. In general, it was found that the anisotropic composites were effective in reducing cooling loads; more investigation on heating loads is needed. During FY 18, ORNL performed a series of laboratory tests of anisotropic composites and appropriate heat sink/source, validated the FEMs used for the FY17 numerical study using experimental data, and revisited the whole-building modeling. Wall prototypes were created and tested in ORNL’s large scale climate simulator (LSCS) under variable conditions approximating summer and winter time periods.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗
DOE OSTI2021

Structure-preserving numerical discretizations for domains with boundaries

This SAND report documents Exploratory Express LDRD Project 223790, "Structure-preserving numerical discretizations for domains with boundaries", which developed a method to incorporate consistent treatment of domain boundaries and arbitrary boundary conditions in discrete exterior calculus (DEC) for arbitrary polygonal (2D) and tensor-product structure prism (3D) grids. The new DEC required the development of novel discrete exterior derivatives, boundary operators, wedge products and Hodge stars. This was accomplished through the use of boundary extension and the blending of known 2D operators on the interior with 1D operators on the boundary. The Hodge star was based on the Voronoi Hodge star, and retained the limitation of a triangular circumcentric primal or dual grid along with low-order accuracy. In addition to the new DEC, two related software packages were written: one for the study of DEC operators on arbitrary polygonal and polyhedral grids using both symbolic and numerical approaches and one for a (thermal) shallow water testbed using TRiSK-type numerics. Immediately relevant (already funded, through CANGA) followup work is the development of a high-order, geometrically flexible Hodge star and structure-preserving, high-order, oscillation-limiting transport operators (using WENO) for n-forms on arbitrary 2D and 3D grids. This will provide all of the machinery required for a high-order version of TRiSK with boundaries on arbitrary 2D and tensor-product 3D grids, which is applicable to both the atmospheric (CRM in E3SM-MMF) and oceanic (MPAS-O) components of E3SM.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2021

Final report of activities for the LDRD-express project #223796 titled: “Fluid models of charged species transport: numerical methods with mathematically guaranteed properties”, PI: Ignacio Tomas, Co-PI: John Shadid

This report summarizes the findings and outcomes of the LDRD-express project with title “Fluid models of charged species transport: numerical methods with mathematically guaranteed properties”. The primary motivation of this project was the computational/mathematical exploration of the ideas advanced aiming to improve the state-of-the-art on numerical methods for the one-fluid Euler-Poisson models and gain some understanding on the Euler-Maxwell model. Euler-Poisson and Euler-Maxwell, by themselves are not the most technically relevant PDE plasma-models. However, both of them are elementary building blocks of PDE-models used in actual technical applications and include most (if not all) of their mathematical difficulties. Outside the classical ideal MHD models, rigorous mathematical and numerical understanding of one-fluid models is still a quite undeveloped research area, and the treatment/understanding of boundary conditions is minimal (borderline non-existent) at this point in time. This report focuses primarily on bulk-behaviour of Euler-Poisson’s model, touching boundary conditions only tangentially.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2024

Numerical Simulation of Infrasound Resonance in Underground Tunnel Structures

Remote observation of infrasound resonant signals emanating from underground tunnel structures could potentially allow the remote quantification of the geometry of the underground structures in which the signals were generated. However, the sensitivity of these observations to tunnel geometry and changes in that geometry are unknown. In this report we outline a numerical simulation study with the following three objectives: 1. Can we model infrasound resonance using numerical simulations? 2. What is the sensitivity of the modeled observations to changes in tunnel geometry or boundary conditions? And 3. Can we accurately model resonant observations of explosions occurring in Redmond Salt Mine in Central Utah. In this report we outline affirmative answers to the first two objectives, but we were unable to accurately model the Redmond Explosions due to numerical instability in the model of the complex structure of the mine. Recommendations for future work emphasize the need to acquire additional datasets and to explore a more data-based approach in which changes in data signatures are detected as a first step towards developing a method which inverts resonant infrasound signals for tunnel geometry.

47 OTHER INSTRUMENTATION↗
DOE OSTI2025

Validation of Numerical Tools for Calculating Reactivity Feedback in Sodium Fast Reactors Using SEFOR Experimental Data

The Southwest Experimental Fast Oxide Reactor (SEFOR) was an experimental sodium-cooled fast breeder reactor operated from 1969 to 1972 with experiments designed to measure Doppler reactivity feedback in a wide temperature range from around 350 °F to temperatures approaching the melting point of mixed oxide fuel of around 5000 °F, providing valuable data for code validations. Co-supported by the Department of Energy (DOE) Fast Reactor Program (FRP) and the DOE Nuclear Energy Advanced Modeling and Simulation (NEAMS) program, the SEFOR benchmark project focused on using the experimental data to validate numerical tools that are used in industry and academia to design and license sodium-cooled fast reactors (SFRs). By the end of FY-25, substantial progress was achieved in the SEFOR benchmark study. A variety of numerical tools commonly used for modeling SFRs were applied to develop models for SEFOR core configurations I-D, I-E, I-I, and I-J. These included Monte Carlo codes such as MCNP, Serpent, and Shift; deterministic codes such as the legacy Argonne Reactor Computation (ARC) suite and the high-fidelity NEAMS code Griffin; and the system analysis code SAS4A/SASSYS-1 (SAS). Using these models, both SEFOR zero-power experiments and power-ascending tests were successfully simulated. Comparisons were performed against experimental measurements of core criticalities, reflector worth, kinetics parameters (Λ/βeff), isothermal reactivity feedback (from 350 °F to 760 °F at zero power), and power-ascending reactivity feedback (as power increased from 0.4 MW to 17 MW). In general, these comparisons demonstrated very good agreement between numerical results and experimental data. In Fiscal Year 26 (FY-26), the SEFOR benchmark project will continue to address the modeling issues identified in FY-25. Effort will focus on the simulation of reactivity insertion transients in SEFOR core II using the ARC/SAS model. Future work will also focus on incorporating BISON into the SEFOR core modeling process to enable the first Multiphysics simulations of the isothermal tests based on the MOOSE framework.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗
DOE OSTI2026

Numerical Investigation of Fluid Flow and Space Charge in Liquid Argon Time Projection Chamber (LArTPC) Detectors

Overview This project focused on developing a high-fidelity numerical framework to simulate the multiphysics environment within Liquid Argon Time Projection Chamber (LArTPC) detectors. The primary objective was to characterize the complex interplay between ion transport, background fluid dynamics, and electric field distortions—a critical factor for the calibration and sensitivity of next-generation High Energy Physics experiments, such as DUNE. Technical Achievements The research successfully yielded a hybrid numerical space-charge solver utilizing a Cell-Centered Finite Volume Method (FVM) for ion transport coupled with a Finite Element Method (FEM) for electric potential. Key accomplishments include: • Verification & Validation: The 3-D solver was rigorously verified against 1-D analytical solutions, demonstrating high numerical accuracy in predicting space-charge-induced field deviations. • Field Distortion Analysis: 3D simulations revealed that space charge effects introduce significant non-uniformities in the electric field. Critically, the research identified that background LAr flow velocities, when comparable to ion drift velocities, markedly exacerbate these distortions. • Technology Transfer: The resulting source code and comprehensive user manuals were successfully transferred to collaborators at Fermilab, providing a portable computational tool for the broader scientific community. Challenges and Future Directions While the space-charge solver achieved all performance metrics, the integrated fluid dynamics modeling encountered convergence challenges stemming from the extreme 200-fold disparity in length scales between the detector's 37 mm inlet pipes and the 8-meter global domain. To address this, the project has identified a clear technical pivot toward Hierarchical Geometric Adaptive Mesh Refinement (HG-AMR). By implementing an h-type refinement strategy with hanging nodes, future iterations of this solver will be capable of resolving localized high-gradient inlet flows without the prohibitive computational costs of regular grids. This advancement, combined with data-driven uncertainty quantification based on MicroBooNE-style calibration, will enable the precise modeling of detector responses in large-scale cryogenic environments where direct measurement remains difficult. Impact The computational tools developed under this award provide a foundation for enhancing the energy resolution and spatial reconstruction of noble liquid detectors. By bridging the gap between theoretical fluid dynamics and experimental field calibration, this work supports the DOE’s mission to advance the frontiers of neutrino physics and dark matter detection.

42 ENGINEERING↗
DOE OSTI2016

Numerical Benchmark for High-Reynolds-Number Supercritical Flows with Large Density Gradients

Because of the extreme complexity of physical phenomena at high pressure, only limited data are available for solver validation at device-relevant conditions such as liquid rocket engines, gas turbines, or diesel engines. In the present study, a two-dimensional direct numerical simulation is used to establish a benchmark for supercritical flow at a high Reynolds number and high-density ratio at conditions typically encountered in liquid rocket engines. Emphasis has been placed on maintaining the flow characteristics of actual systems with simple boundary conditions, grid spacing, and geometry. Results from two different state-of-the-art codes, with markedly different numerical formalisms, are compared using this benchmark. The strong similarity between the two numerical predictions lends confidence to the physical accuracy of the results. The established database can be used for solver benchmarking and model development at conditions relevant to many propulsion and power systems.

Engineering↗
DOE OSTI2021

NUMERICAL CALCULATION OF LOSSES OF TRAPPED VORTICES UNDER STRONG RF MEISSNER CURRENT AND DC SUPERHEATING FIELD IN TYPE II SUPERCONDUCTORS

Research on the vortex dynamics and enhancing of superheating field in superconductors has attracted much attention in accelerator physics community to develop next-generation high-performance accelerator cavities. However, the extreme dynamics of curvilinear elastic vortices driven by very strong currents close to the depairing limit or superheating field of a superconductor with a nanostructured surface has not been well understood. We calculated the superheating field Hsh and critical momentum kc characterizing the wavelength of the instability ?m of the Meissner state to flux penetration by solving numerically the Ginzburg?Landau equations. A bulk superconductor, superconductor with the inhomogeneous surface disorder (S-S), and multilayered surface (S-I-S) have been thoroughly investigated in this work. Our result showed that S-S and S-I-S structures can enhance the superheating field well above their clean limit. In this work extensive numerical simulation of the power dissipated by an oscillating vortex segment driven by the surface ac Meissner currents was performed. Our simulations take into account the nonlinear vortex line tension, vortex mass, Bardeen?Stephen viscous vortex drag applicable at low fields, and nonlinear Larkin-Ovchinnikov (LO) viscous drag coefficient ?(v) at high fields and pinning force. We showed that the LO decrease of ?(v) with the vortex velocity v could radically change the field dependence of the surface resistance Ri(H) caused by trapped vortices. At low frequencies Ri(H) exhibits a conventional increase with H. However, as frequency increases, the surface resistance becomes a nonmonotonic function of H which decreases with H at higher fields irrespective of the pinning distribution. Overheating can mask the descending field dependence of Ri(H) as frequency increases. Our numerical simulations also show that the LO effect can cause a vortex bending instability at high field amplitudes and frequencies, giving rise to the formation of dynamic kinks along with the vortex when a vortex is pinned strongly to one end. Nonlinear losses of trapped vortices in thick films under high-amplitude RF fields as functions of frequency, mean free path and pinning characteristics have been calculated.

Pathirana, Walive↗
DOE OSTI2021

Investigation of Scale-Dependent Groundwater/Surface-water Exchange in Rivers by Gradient Self-Potential Logging: Numerical Modeling and Field Experiments

Exchanges of groundwater and surface-water are fundamental to a wide range of water-supply and water-quality management issues but challenging to map beyond the reach scale. Waterborne gradient self-potential (SP) measurements are directly sensitive to water flow through riverbed sediments and can be used to infer exchange locations, direction (gain versus loss), scale, and relative changes, but to date applications to river corridor hydrology are limited. Numerical modeling and field experiments were therefore performed herein, each emphasizing waterborne gradient SP logging for identifying and locating focused vertical groundwater discharge (surface-water gain) and recharge (surface-water loss) in a river. Two and three-dimensional numerical models were constructed to simulate the polarities, appearances, and peak amplitudes of streaming-potential and electric-field anomalies on a riverbed and in the surface-water that were attributable to steady-state vertical fluxes of groundwater through high-permeability conduits in the riverbed. Effects of varied hydraulic length-scale of exchange and surface-water depth were tested through numerical modeling. Modeling results aided in data acquisition and interpretation for three repeated field experiments performed along a 1.5–2.0 km reach of the Quashnet River in Cape Cod, Massachusetts, where focused, meter-scale groundwater discharges occur at discrete locations within otherwise ubiquitous and more diffuse groundwater upwelling conditions. Strong gradient SP anomalies were repeatedly measured in the Quashnet River at previously confirmed locations of focused groundwater discharge, showing the efficacy of waterborne gradient SP logging in identifying and characterizing groundwater/surface water exchange dynamics at multiple river network scales.

Engineering↗
DOE OSTI2022

Numerical investigation of the effect of ultrasound on paper drying

The paper drying process is very energy inefficient. More than two-thirds of the total energy used in a paper machine is for drying paper. Novel drying technologies, such as ultrasound (US) drying, can be assessed numerically for developing next-generation drying technologies for the paper industry.This work numerically illustrates the impact on drying process energy efficiency of US transducers installed on a two-tiered dryer section of a paper machine. Piezoelectric transducers generate ultrasound waves, and liquid water mist can be ejected from the porous media. The drying rate of handsheet paper in the presence of direct-contact USis measured experimentally, and the resultant correlation is included in the theoretical model.The drying section of a paper machine is simulated by a theoretical drying model. In the model, three scenarios are considered. In the first scenario, the US modules are positioned in the dryer pockets, while in the second scenario, they are placed upstream of the drying section right after the press section. The third case is the combination ofthe first and second scenarios. The average moisture content and temperature during drying, enhancement of total mass flux leaving the paper by the US mechanism, total energy consumption, and thermal effect of heated US transducers are analyzed for all cases. Results show that the application of the US can decrease the total number of dryer drums for drying paper. This numerical study is based on the US correlation obtained with the US transducer in direct contact with the paper sample. Thus, future work should include US correlation based on a non-contact US transducer.

Materials Science↗
DOE OSTI2020

Comparison of Numerical Methods for Modeling the Wave Field Effects Generated by Individual Wave Energy Converters and Multiple Converter Wave Farms

This numerical study compares the wave field generated by the spectral wave action balance code, SNL-SWAN, to the linear-wave boundary-element method (BEM) code, WAMIT. The objective of this study is to assess the performance of SNL-SWAN for modeling wave field effects produced by individual wave energy converters (WECs) and wave farms comprising multiple WECs by comparing results from SNL-SWAN with those produced by the BEM code WAMIT. BEM codes better model the physics of wave-body interactions and thus simulate a more accurate near-field wave field than spectral codes. In SNL-SWAN, the wave field’s energy extraction is modeled parametrically based on the WEC’s power curve. The comparison between SNL-SWAN and WAMIT is made over a range of incident wave conditions, including short-, medium-, and long-wavelength waves with various amounts of directional spreading, and for three WEC archetypes: a point absorber (PA), a pitching flap (PF) terminator, and a hinged raft (HR) attenuator. Individual WECs and wave farms of five WECs in various configuration were studied with qualitative comparisons made of wave height and spectra at specific locations, and quantitative comparisons of the wave fields over circular arcs around the WECs as a function of radial distance. Results from this numerical study demonstrate that in the near-field, the difference between SNL-SWAN and WAMIT is relatively large (between 20% and 50%), but in the far-field from the array the differences are minimal (between 1% and 5%). The resultant wave field generated by the two different numerical approaches is highly dependent on parameters such as: directional wave spreading, wave reflection or scattering, and the WEC’s power curve.

16 TIDAL AND WAVE POWER↗
DOE OSTI2016

A NUMERICAL SCHEME FOR SPECIAL RELATIVISTIC RADIATION MAGNETOHYDRODYNAMICS BASED ON SOLVING THE TIME-DEPENDENT RADIATIVE TRANSFER EQUATION

We develop a numerical scheme for solving the equations of fully special relativistic, radiation magnetohydrodynamics (MHDs), in which the frequency-integrated, time-dependent radiation transfer equation is solved to calculate the specific intensity. The radiation energy density, the radiation flux, and the radiation stress tensor are obtained by the angular quadrature of the intensity. In the present method, conservation of total mass, momentum, and energy of the radiation magnetofluids is guaranteed. We treat not only the isotropic scattering but also the Thomson scattering. The numerical method of MHDs is the same as that of our previous work. The advection terms are explicitly solved, and the source terms, which describe the gas–radiation interaction, are implicitly integrated. Our code is suitable for massive parallel computing. We present that our code shows reasonable results in some numerical tests for propagating radiation and radiation hydrodynamics. Particularly, the correct solution is given even in the optically very thin or moderately thin regimes, and the special relativistic effects are nicely reproduced.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2016

THE FORMATION OF A MILKY WAY-SIZED DISK GALAXY. I. A COMPARISON OF NUMERICAL METHODS

The long-standing challenge of creating a Milky Way- (MW-) like disk galaxy from cosmological simulations has motivated significant developments in both numerical methods and physical models. We investigate these two fundamental aspects in a new comparison project using a set of cosmological hydrodynamic simulations of an MW-sized galaxy. In this study, we focus on the comparison of two particle-based hydrodynamics methods: an improved smoothed particle hydrodynamics (SPH) code Gadget, and a Lagrangian Meshless Finite-Mass (MFM) code Gizmo. All the simulations in this paper use the same initial conditions and physical models, which include star formation, “energy-driven” outflows, metal-dependent cooling, stellar evolution, and metal enrichment. We find that both numerical schemes produce a late-type galaxy with extended gaseous and stellar disks. However, notable differences are present in a wide range of galaxy properties and their evolution, including star-formation history, gas content, disk structure, and kinematics. Compared to Gizmo, the Gadget simulation produced a larger fraction of cold, dense gas at high redshift which fuels rapid star formation and results in a higher stellar mass by 20% and a lower gas fraction by 10% at z = 0, and the resulting gas disk is smoother and more coherent in rotation due to damping of turbulent motion by the numerical viscosity in SPH, in contrast to the Gizmo simulation, which shows a more prominent spiral structure. Given its better convergence properties and lower computational cost, we argue that the MFM method is a promising alternative to SPH in cosmological hydrodynamic simulations.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2017

Two-fluid Numerical Simulations of Solar Spicules

We aim to study the formation and evolution of solar spicules by means of numerical simulations of the solar atmosphere. With the use of newly developed JOANNA code, we numerically solve two-fluid (for ions + electrons and neutrals) equations in 2D Cartesian geometry. We follow the evolution of a spicule triggered by the time-dependent signal in ion and neutral components of gas pressure launched in the upper chromosphere. We use the potential magnetic field, which evolves self-consistently, but mainly plays a passive role in the dynamics. Our numerical results reveal that the signal is steepened into a shock that propagates upward into the corona. The chromospheric cold and dense plasma lags behind this shock and rises into the corona with a mean speed of 20–25 km s{sup −1}. The formed spicule exhibits the upflow/downfall of plasma during its total lifetime of around 3–4 minutes, and it follows the typical characteristics of a classical spicule, which is modeled by magnetohydrodynamics. The simulated spicule consists of a dense and cold core that is dominated by neutrals. The general dynamics of ion and neutral spicules are very similar to each other. Minor differences in those dynamics result in different widths of both spicules with increasing rarefaction of the ion spicule in time.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2013

A numerical study of the relationship between NMR relaxation and permeability in sands and gravels

The accuracy of NMR-derived permeability estimates in sands and gravels were examined through simulations on numerical grain packs composed of uniform spherical grains. The packs consisted of randomly packed grains, with grain sizes set to represent a range corresponding to sands and gravels. The material properties for each pack were quantified through numerical analysis and the NMR response was simulated for a range of surface relaxivity values. The agreement between the numerically-derived permeability estimates and the permeability estimates derived using the Schlumberger Doll Research (SDR) and Seevers equations was evaluated. Use of the SDR equation assumes that the relaxation of the bulk pore fluid can be neglected. The NMR-derived permeability estimates were calculated using each equation for the cases where relaxation was assumed to occur in one of the two major diffusion regimes. We found that permeability is most accurately estimated in all packs through use of the Seevers equation with the empirical constant n set equal to 1. We showed that the contribution of bulk fluid relaxation should be accounted for in materials with grain radii greater than 1.2e-4m (fine sand) and surface relaxivity values less than 1.0e-3 m s -1 20 . In practice, this range of surface relaxivity values and grain sizes corresponds to situations where the measured relaxation time T 2 is greater than approximately one-third the value of the bulk fluid relaxation time T 2B .

Dlubac, Katherine↗
DOE OSTI2018

Numerical Methodology for Optimization of Compression-Ignited Engines Considering Combustion Noise Control

It is challenging to develop highly efficient and clean engines while meeting user expectations in terms of performance, comfort, and drivability. One of the critical aspects in this regard is combustion noise control. Combustion noise accounts for about 40 percent of the overall engine noise in typical turbocharged diesel engines. The experimental investigation of noise generation is difficult due to its inherent complexity and measurement limitations. Therefore, it is important to develop efficient numerical strategies in order to gain a better understanding of the combustion noise mechanisms. In this work, a novel methodology was developed, combining computational fluid dynamics (CFD) modeling and genetic algorithm (GA) technique to optimize the combustion system hardware design of a high-speed direct injection (HSDI) diesel engine, with respect to various emissions and performance targets including combustion noise. The CFD model was specifically set up to reproduce the unsteady pressure field inside the combustion chamber, thereby allowing an accurate prediction of the acoustic response of the combustion phenomena. The model was validated by simulating several steady operating conditions and comparing the numerical results against experimental data, in both temporal and frequency domains. Thereafter, a GA optimization was performed with the goal of minimizing indicated specific fuel consumption (ISFC) and combustion noise, while restricting pollutant (soot and NOx) emissions to their respective baseline values. Eight design variables were selected pertaining to piston bowl geometry, nozzle inclusion angle, number of injector nozzle holes, and in-cylinder swirl. An objective merit function (MF) based on the emissions, ISFC, and combustion noise was constructed to quantify the strength of the engine designs and was determined using the CFD model as the function evaluator. The in-cylinder noise level was characterized by the total resonance energy of local pressure oscillations. The optimum engine configuration thus obtained showed a significant improvement in terms of efficiency and combustion noise compared to the baseline system, along with both soot and NOx emissions within their respective constraints. This optimum configuration included a deeper and tighter bowl geometry with higher swirl and larger number of nozzle holes. Subsequently, a more detailed acoustics analysis based on proper orthogonal decomposition (POD) technique was carried out to further explore the combustion noise benefits achieved by the GA optimum. In conclusion, this computational study is a first of its kind (to the best of the authors’ knowledge), which demonstrates a comprehensive framework to incorporate combustion noise into a numerical optimization strategy for engine design.

42 ENGINEERING↗
DOE OSTI2020

Documenting numerical experiments in support of the Coupled Model Intercomparison Project Phase 6 (CMIP6)

Abstract. Numerical simulation, and in particular simulation of the earth system, relies on contributions from diverse communities, from those who develop models to those involved in devising, executing, and analysing numerical experiments. Often these people work in different institutions and may be working with significant separation in time (particularly analysts, who may be working on data produced years earlier), and they typically communicate via published information (whether journal papers, technical notes, or websites). The complexity of the models, experiments, and methodologies, along with the diversity (and sometimes inexact nature) of information sources, can easily lead to misinterpretation of what was actually intended or done. In this paper we introduce a taxonomy of terms for more clearly defining numerical experiments, put it in the context of previous work on experimental ontologies, and describe how we have used it to document the experiments of the sixth phase for the Coupled Model Intercomparison Project (CMIP6). We describe how, through iteration with a range of CMIP6 stakeholders, we rationalized multiple sources of information and improved the clarity of experimental definitions. We demonstrate how this process has added value to CMIP6 itself by (a) helping those devising experiments to be clear about their goals and their implementation, (b) making it easier for those executing experiments to know what is intended, (c) exposing interrelationships between experiments, and (d) making it clearer for third parties (data users) to understand the CMIP6 experiments. We conclude with some lessons learnt and how these may be applied to future CMIP phases as well as other modelling campaigns.

Pascoe, Charlotte (ORCID:0000000223616778)↗
DOE OSTI2024

When ancient numerical demons meet physics-informed machine learning: adjoint-based gradients for implicit differentiable modeling

Recent advances in differentiable modeling, a genre of physics-informed machine learning that trains neural networks (NNs) together with process-based equations, have shown promise in enhancing hydrological models' accuracy, interpretability, and knowledge-discovery potential. Current differentiable models are efficient for NN-based parameter regionalization, but the simple explicit numerical schemes paired with sequential calculations (operator splitting) can incur numerical errors whose impacts on models' representation power and learned parameters are not clear. Implicit schemes, however, cannot rely on automatic differentiation to calculate gradients due to potential issues of gradient vanishing and memory demand. Here we propose a “discretize-then-optimize” adjoint method to enable differentiable implicit numerical schemes for the first time for large-scale hydrological modeling. The adjoint model demonstrates comprehensively improved performance, with Kling–Gupta efficiency coefficients, peak-flow and low-flow metrics, and evapotranspiration that moderately surpass the already-competitive explicit model. Therefore, the previous sequential-calculation approach had a detrimental impact on the model's ability to represent hydrological dynamics. Furthermore, with a structural update that describes capillary rise, the adjoint model can better describe baseflow in arid regions and also produce low flows that outperform even pure machine learning methods such as long short-term memory networks. The adjoint model rectified some parameter distortions but did not alter spatial parameter distributions, demonstrating the robustness of regionalized parameterization. Despite higher computational expenses and modest improvements, the adjoint model's success removes the barrier for complex implicit schemes to enrich differentiable modeling in hydrology.

58 GEOSCIENCES↗