SEARCH · Engineering Papers

Results for “Numerical techniques”

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 991 records · Page 55

All sources NASA NTRS DOE OSTI

DOE OSTI2023

Understanding fundamental effects of biofuel structure on ignition and physical fuel properties

The development of biochemicals and biofuels with advantaged properties of higher combustion efficiency at lower emission levels can be aided by a priori prediction of critical global characteristics based on molecular structure. Existing predictive techniques are incomplete for comprehensive ignition and physical fuel property prediction. Here this investigation focuses on filling this knowledge gap by developing a priori prediction techniques for predicting important ignition as well as physical fuel properties of promising advantaged biofuels. We demonstrate the utility of this a priori prediction technique on saturated (≤C 5 ) alcohols, both linear and branched in nature by correlating their molecular structure with both important ignition and physical properties of the fuels. This work utilizes a semi-automated method of predicting ignition characteristics via accurate fundamental-based calculations of the fuel radical cascades that govern low-temperature combustion. This numerical methodology is supported by the relevant fuel ignition delay time (IDT) and research octane number (RON) data experimentally measured by the Advanced Fuel Ignition Delay Analyzer (AFIDA). Phyiscal fuel property trends are also analyzed by correlating the molecular structure of the fuel to its physical characteristics. Based on their molecular structure, recommendations are made on an a priori method for selecting biofuels that have advantaged properties for the selected fuel application. Lastly, the utility of this investigation is extended beyond the subset of alcohols considered by analyzing the IDT trends of i-Pentanol and n-Hexanol, revealing the fidelity of the novel kinetic modelling approach and the accuracy of the predicted trends in this study.

09 BIOMASS FUELS↗

DOE OSTI2023

Towards Quantum Computing Phase Diagrams of Gauge Theories with Thermal Pure Quantum States

The phase diagram of strong interactions in nature at finite temperature and chemical potential remains largely theoretically unexplored due to inadequacy of Monte-Carlo–based computational techniques in overcoming a sign problem. Quantum computing offers a sign-problem-free approach, but evaluating thermal expectation values is generally resource intensive on quantum computers. To facilitate thermodynamic studies of gauge theories, we propose a generalization of the thermal-pure-quantum-state formulation of statistical mechanics applied to constrained gauge-theory dynamics, and numerically demonstrate that the phase diagram of a simple low-dimensional gauge theory is robustly determined using this approach, including mapping a chiral phase transition in the model at finite temperature and chemical potential. Quantum algorithms, resource requirements, and algorithmic and hardware error analysis are further discussed to motivate future implementations. Thermal pure quantum states, therefore, may present a suitable candidate for efficient thermal simulations of gauge theories in the era of quantum computing.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

DOE OSTI2021

A HPC Theory-Guided Machine Learning Cyberinfrastructure for Communicating Hydrometeorological Data Across Scales

High-resolution predictions of hydrometeorological variables are critical for supporting hydropower generation decisions and flood control at hydroelectric power plants. Traditional climate and hydrologic models rely on the numerical simulation of detailed physical processes. Therefore, running these simulations is time-, labor-, and computation-intensive. Improving the spatial and temporal resolution in these modeling outputs could lead to cubic increases in both the simulation time and computational demands, rendering high-resolution hydrometeorological predictions expensive and impractical. Many past studies apply the super resolution (SR) technique to downscale climate models using deep learners. However, deep learners are deemed “black-boxes,” as their derivation processes from low-resolution outputs to high-resolution outputs are often hidden. Their results are difficult for domain scientists to interpret and validate. Thus, there is a need for an exploratory machine learning approach that can partially integrate domain-specific theory and knowledge into the data-driven mapping process between simulation outputs of different spatial scales. The domain-specific theory and knowledge can be incorporated into the data model through an inductive approach in which process-related environmental variables are used and analyzed as key drivers (i.e., environmental surrogates) to reflect the complex physical processes. Many of these variables, such as land use land cover, soil types, topography, digital elevation, air temperature, and various watershed characteristics, can be directly measured through sensors or remote sensing techniques. Additionally, SR applications that can downscale hydrological and hydrodynamics models to efficiently produce high-resolution (1 m) flood depth grids are still rare. Since the flood depth grid can be used to support critical decisions for flood control operation at hydroelectric power plants, it is crucial to enable an SR-based capability for interpolating high-resolution flood inundation maps.

13 HYDRO ENERGY↗

DOE OSTI2025

Prediction and Analysis of Utah FORGE Injection Activities using a Coupled Thermo-hydro-mechanical and Earthquake (THM+E) Modeling Workflow

A coupled thermo-hydro-mechanical (THM) numerical workflow that is capable of modeling seismic slip is critical for the successful development of enhanced geothermal systems (EGS). By integrating key physical processes, this workflow enables accurate simulation of temperature and pressure diffusions, stress changes, and induced seismicity. As a result, it serves as a vital tool for predicting induced seismicity and optimizing reservoir stimulation strategies. The Utah FORGE (Frontier Observatory for Research in Geothermal Energy) project, located near Milford, Utah, is a U.S. Department of Energy initiative aimed at advancing EGS technology. In April 2024, eight new stimulation stages (Stages 3R-10) were conducted in well 16A (injection well) subsequent to the first series of stimulation (Stages 1-3) performed in April, 2022. To monitor the induced seismicity, geophones were deployed in wells 58-32, 56-32, and 78B-32, while fiber optic cables were also installed in wells 16B, 78-32, and 78B-32 to collect microseismic data and detect frac hits Preliminary analyses of microseismic catalogs and fiber optic data suggest that the stimulated fractures in Stages 3R–6 closely align with that generated during Stage 3, indicating that the new stimulations were likely reactivating the previously stimulated fracture. To better understand the underlying process, a comprehensive modeling approach that can accurately capture thermal, hydrological, mechanical, and seismic responses is essential. In this work, we propose and utilize a coupled thermo-hydro-mechanical and earthquake (THM+E) simulation workflow to numerically investigate the stimulation activities on well 16A. The specific objective is to confirm whether the new stimulation stages (Stages 3R–6) reactivated fractures previously stimulated during Stage 3. For this purpose, we perform THM+E simulations individually for Stages 3, 3R, 4, and 5, incorporating the discrete fracture networks (DFNs) created by the plane-fitting technique based on the microseismic catalogs. The simulation workflow consists of two separate models: a THM model and an earthquake model, coupled in a one-way manner. Detailed descriptions of the workflow are provided in Section 3. Simulation results are presented in terms of injection pressure, permeability evolution, and predicted seismic catalogs, which are then compared with field data for further analyses. This report is structured as follows. In Section 2, we present detailed analyses of the field data and propose the hypothesis that the new stimulation stages (Stages 3R–6) were probably reactivating the previously stimulated fractures in Stage 3. In Section 3, we introduce the coupled THM+E workflow and the problem setup to validate our hypothesis, followed by the simulation results for each stage in Section 4. Meanwhile, discussions are included to analyze the model predictions and their comparison with field data. Lastly, we conclude the report and outline future plans in Section 5.

15 GEOTHERMAL ENERGY↗

DOE OSTI2023

A Mixed integer linear programming‐based distributed energy management for networked microgrids considering network operational objectives and constraints

Abstract Mixed integer linear programming (MILP)–based distributed energy management for networked microgrids embedded modern distribution systems is proposed. Considering the diverse ownership of microgrids, distributed energy resources (DERs) that interface directly with utilities and responsive loads, an alternating direction method of multipliers–based distributed framework was formulated for the scheduling of networked microgrids embedded modern distribution systems by adjusting nodal price signals iteratively. In addition, to make the formulated optimization problems resolvable through more accessible and popular MILP solvers, different linearisation techniques were employed to transform the nonlinear terms into linear or mixed integer linear formats. The proposed MILP‐based distributed method preserves all participants' autonomy (e.g., microgrids, DERs that interface directly with utilities and responsive loads), while incentivising them to actively participate in the distribution system operation with price signals. The proposed method is validated with results of numerical simulation using a modern distribution system consisting of multiple networked microgrids, DERs that interface directly with utilities, as well as responsive loads.

24 POWER TRANSMISSION AND DISTRIBUTION↗

DOE OSTI2025

Laser-initiated p – 11 B fusion reactions in petawatt high-repetition-rate laser facilities

Driving of the nuclear fusion reaction p + 11 B → 3α + 8.7 MeV under laboratory conditions by interaction between high-power laser pulses and matter has become a popular field of research, owing to its numerous potential applications: as an alternative to deuterium–tritium for fusion energy production, astrophysics studies, and alpha-particle generation for medical treatment. One possible scheme for laser-driven p– 11 B reactions is to direct a beam of laser-accelerated protons onto a boron (B) sample (the so-called “pitcher-catcher” scheme). This technique has been successfully implemented on large high-energy lasers, yielding hundreds of joules per shot at low repetition. We present here a complementary approach, exploiting the high repetition rate of the VEGA III petawatt laser at CLPU (Spain), aiming at accumulating results from many interactions at much lower energy, to provide better control of the parameters and the statistics of the measurements. Despite a moderate energy per pulse, our experiment allowed exploration of the laser-driven fusion process with tens (up to hundreds) of laser shots. The experiment provided a clear signature of the reactions involved and of the fusion products, accumulated over many shots, leading to an improved optimization of the diagnostics for experimental campaigns of this type. In this paper, we discuss the effectiveness of laser-driven p– 11 B fusion in the pitcher–catcher scheme, at a high repetition rate, addressing the challenges of this experimental scheme and highlighting its critical aspects. Our proposed methodology allows evaluation of the performance of this scheme for laser-driven alpha particle production and can be adapted to high-repetition-rate laser facilities with higher energy and intensity.

Alpha particles↗

DOE OSTI2025

A Standardized Analysis Process Using Digital Image Correlation to Calculate In Situ Cladding Strain from Modified Burst Tests for Fuel Performance Code Validation

Historical data collection on nuclear fuel cladding materials has focused on generating a statistically significant amount of data to assess the material and its failure behavior. Furthermore, data generated to support material model and failure criteria development were previously posttest evaluations, so a large number of tests was required to gain new understanding. A way to expedite this process is to develop techniques capable of generating large, high-fidelity data sets from a single test with lower uncertainty or quantified uncertainty. One such example of this approach is Oak Ridge National Laboratory’s use of modified burst tests (MBTs) to analyze the mechanical behavior and failure conditions of cladding during a simulated reactivity-initiated accident (RIA). Each test incorporates digital image correlation (DIC) analysis techniques that are used to assess the accumulated strain in situ, as well as eventual cladding failure. This work has been fruitful in defining strain-to-failure conditions for materials like silicon carbide (SiC) fiber–reinforced/SiC matrix composite tubes (SiC/SiC), iron-chromium-aluminum (FeCrAl) alloy tubes, and chromium-coated Zircaloy-4 tubes. However, there are numerous DIC software available, including open-source and proprietary software. The different DIC software use various algorithms to process images and calculate displacement values. Using these different software and algorithms can lead to varying results, and perhaps larger-than-expected uncertainties. In the present study, previously published MBT data encompassing a variety of test conditions were reanalyzed with two different DIC software to assess the variance in the calculated strain results. The data consisted of SiC/SiC, FeCrAl, and chromium-coated Zircaloy-4 tubes. Plots of the calculated strains during the transient revealed good agreement between the two DIC software. The average root-mean-square errors between the two software was 0.20% strain, which is slightly larger than a previously reported error value for these tests. In conclusion, this variance in results is low enough that this analysis method can be used for code validation.

Reactivity-initiated accident↗

DOE OSTI2021

Comparison of Rotary and Linear Cutting Methodology in Determining Specific Cutting Energy of Granite

Abstract Single polycrystalline diamond compact (PDC) cutting is a practical technique to understand the rock–tool interactions in drag-bit-type geothermal drilling operations. This paper introduces a rotary cutting method to determine specific cutting energy (SCE) and compares it with the conventional linear cutting method. In this work, granite is selected to represent hard rock formations in geothermal drilling. Cutting tests are conducted on a Computer Numerical Control (CNC) machine with a realistic cutting speed of 12.7 m/min and several chip loads ranging from 0.08 to 0.25 mm. The cutting force is measured using a dynamometer and then converted to specific cutting energy (SCE). The results show that the rotary method produces an inverse relationship between SCE and chip load, whereas the linear method shows the opposite. As a result, the produced SCE by the rotary method tends to be lower than that of the linear method at a higher chip load at and over 0.16 mm. The difference may be attributed to the cutting configuration and associated force components.

Engineering↗

DOE OSTI2020

Effect of heating/cooling dynamics in the hysteresis loop and tunable IR emissivity of VO 2 thin films

We experimentally investigate the semiconductor-to-metal transition (SMT) in vanadium dioxide thin films using an infrared thermographic technique. During the semiconductor to metal phase change process, VO 2 optical properties dynamically change and infrared emission undergoes a hysteresis loop due to differences between heating and cooling stages. The shape of the hysteresis loop was accurately monitored under different dynamic heating/cooling rates. In order to quantify and understand the effects of different rates, we used a numerical modelling approach in which a VO 2 thin layer was modeled as metamaterial. The main experimental findings are interpreted assuming that both the rate of formation and shape of metallic inclusions are tuned with the heating/cooling rate. The structural transition from monoclinic to tetragonal phases is the main mechanism for controlling the global properties of the phase transition. However, our experimental results reveal that the dynamics of the heating/cooling process can become a useful parameter for further tuning options and lays out a macroscopic optical sensing scheme for the microscopic phase change dynamics of VO 2 . Our study sheds light on phase-transition dynamics and their effect on the infrared emission spectra of VO 2 thin films, therefore enabling the heating/cooling rate to be an additional parameter to control infrared emission characteristics of thermal emitters. The hysteresis loop represents the phase coexistence region, thus being of fundamental importance for several applications, such as the operation of radiative thermal logic elements based on phase transition materials. For such applications, the phase transition region is shifted for heating and cooling processes. We also show that, depending on the way the phase change elements are heated, the temperature operation range will be slightly modified.

42 ENGINEERING↗

DOE OSTI2021

Initial conditions for simulations of beam physics in linear induction accelerators

Flash radiography of hydrodynamic experiments driven by high explosives is a well-known diagnostic technique in use at many laboratories. At Los Alamos, the DualAxis Radiographic Hydrodynamic Test (DARHT) facility two linear induction electron accelerators (LIAs) make the bremsstrahlung radiographic source spots for point projection radiographs from orthogonal views. A new LIA, called Scorpius, is presently under development to advance this technology. To better understand electron-beam physics in these LIAs, numerical simulations are frequently performed with the objective of improving the radiography. At Los Alamos we frequently use the TRAK ray-trace and LSP particle-in-cell (PIC) codes to simulate the injector, and the XTR and LAMDA envelope/centroid codes along with LSP to simulate transport of the accelerated beam through the LIAs. The LIA simulations need the injected beam parameters as initial conditions for calculating beam transport and stability. The determination of these initial conditions is the topic of this note.

43 PARTICLE ACCELERATORS↗

DOE OSTI2021

Analytical solutions of the Arrhenius-Semenov problem for constant volume burn

Analytical solutions to the Semenov thermal ignition problem for constant volume burn governed by Arrhenius reaction kinetics are derived. Specifically, an approximate analytical solution technique for the Arrhenius-Semenov differential equation is derived for reaction orders n ϵ R> 0 and exact solutions are also constructed for reaction orders n ϵ N : n ≤ 3. The approximation technique relies on expansion of the respective nondominant terms in the differential equation at the lower and upper bounds of the reaction progress variable in order to create a pair of integrable series. The two integrated series are then connected to create a single continuous analytical solution. Excellent agreement is observed between the analytical approximation and solutions obtained numerically. The presented approximation constitutes a simple and robust strategy for solving the Arrhenius-Semenov problem analytically.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

DOE OSTI2023

Analysis of Fire-Induced Circulations during the FireFlux2 Experiment

Despite recent advances in both coupled fire modeling and measurement techniques to sample the fire environment, the fire–atmosphere coupling mechanisms that lead to fast propagating wildfires remain poorly understood. This knowledge gap adversely affects fire management when wildland fires propagate unexpectedly rapidly and shift direction due to the fire impacts on local wind conditions. In this work, we utilized observational data from the FireFlux2 prescribed burn and numerical simulations performed with a coupled fire–atmosphere model WRF-SFIRE to assess the small-scale impacts of fire on local micrometeorology under moderate wind conditions (10–12 m/s). The FireFlux2 prescribed burn provided a comprehensive observational dataset with in situ meteorological observations as well as IR measurements of fire progression. To directly quantify the effects of fire–atmosphere interactions, two WRF-SFIRE simulations were executed. One simulation was run in a two-way coupled mode in which the heat and moisture fluxes emitted from the fire were injected into the atmosphere, and the other simulation was performed in a one-way coupled mode for which the atmosphere was not affected by the fire. The difference between these two simulations was used to analyze and quantify the fire impacts on the atmospheric circulation at different sections of the fire front. The fire-released heat fluxes resulted in vertical velocities as high as 10.8 m/s at the highest measurement level (20 m above ground level) gradually diminishing with height and dropping to 7.9 m/s at 5.77 m. The fire-induced horizontal winds indicated the strongest fire-induced flow at the lowest measurement levels (as high as 3.3 m/s) gradually decreasing to less than 1 m/s at 20 m above ground level. The analysis of the simulated flow indicates significant differences between the fire-induced circulation at the fire head and on the flanks. The fire-induced circulation was much stronger near the fire head than at the flanks, where the fire did not produce particularly strong cross-fire flow and did not significantly change the lateral fire progression. However, at the head of the fire the fire-induced winds blowing across the front were the strongest and significantly accelerated fire progression. The two-way coupled simulation including the fire-induced winds produced 36.2% faster fire propagation than the one-way coupled run, and more realistically represented the fire progression.

Benik, Jeremy T.↗

DOE OSTI2022

PAS: Privacy Algorithms in Systems

Today we face an explosion of data generation, ranging from health monitoring to national security infrastructure systems. More and more systems are connected to the Internet that collects data at regular time intervals. These systems share data and use machine learning methods for intelligent decisions, which resulted in numerous real-world applications (e.g., autonomous vehicles, recommendation systems, and heart-rate monitoring) that have benefited from it. However, these approaches are prone to identity thief and other privacy related cyber-security attacks. So, how can data privacy be protected efficiently in these scenarios? More dedicated efforts are needed to propose the integration of privacy techniques into existing systems and develop more advanced privacy techniques to address the complex challenges of multi-system connectivity and data fusion. Therefore, we have introduced Privacy Algorithms in Systems (PAS) at CIKM which provides a venue to gather academic researchers and industry researchers/practitioners to present their research in an effort to advance the frontier of this critical direction of privacy algorithms in systems.

Kotevska, Olivera↗

DOE OSTI2024

Local step-flow dynamics in thin film growth with desorption

Desorption of deposited species plays a role in determining the evolution of surface morphology during crystal growth when the desorption time constant is short compared with the time to diffuse to a defect site, step edge, or kink. However, experiments to directly test the predictions of these effects are lacking. Novel techniques such as in situ coherent x-ray scattering can provide significant new information. Herein we present x-ray photon correlation spectroscopy (XPCS) measurements during diindenoperylene (DIP) vapor deposition on thermally oxidized silicon surfaces. DIP forms a nearly complete two-dimensional first layer over the range of temperatures studied (40–120 ⁢°C), followed by mounded growth during subsequent deposition. Local step flow within mounds was observed, and we find that there was a terrace-length-dependent behavior of the step edge dynamics. This led to unstable growth with rapid roughening (β>0.5) and deviation from a symmetric error-function-like height profile. At high temperatures, the grooves between the mounds tend to close up leading to nearly flat polycrystalline films. In conclusion, numerical analysis based on a (1+1)-dimensional model suggests that terrace-length dependent desorption of deposited ad-molecules is an essential cause of the step dynamics, and it influences the morphology evolution.

36 MATERIALS SCIENCE↗

DOE OSTI2021

A review of modelling techniques for floating offshore wind turbines

Modelling floating offshore wind turbines (FOWTs) is challenging due to the strong coupling between the aerodynamics of the turbine and the hydrodynamics of the floating platform. Physical testing at scale is faced with the additional challenge of the scaling mismatch between Froude number and Reynolds number due to working in the two fluid domains, air and water. In the drive for cost-reduction of floating wind energy, designers may be seeking to move towards high-fidelity numerical modelling as a substitute for physical testing. However, the numerical engineering tools typically used for FOWT modelling are considered as mid-fidelity to low-fidelity tools, and currently lack the level of accuracy required to do so. Furthermore, there is a lack of operational FOWT data available for further development and validation. High-fidelity tools, such as CFD, have greater accuracy but are cumbersome tools and still require validation. Physical scale model testing therefore continues to play an essential role in the development of FOWTs both as a source of validation data for numerical models and as an important development step along the path to commercialization of all platform concepts. The aim of this paper is to provide an overview of both numerical modelling and physical FOWT scale model testing approaches and to provide guidance on the selection of the most appropriate approach (or combination of approaches). The current state-of-the-art will be discussed along with current research trends and areas for further investigation.

17 WIND ENERGY↗

DOE OSTI2023

Scalable in situ non-destructive evaluation of additively manufactured components using process monitoring, sensor fusion, and machine learning

Laser Powder Bed Fusion (L-PBF) Additive Manufacturing (AM) is among the metal 3D printing technologies most broadly adopted by the manufacturing industry. However, the current industry qualification paradigm for critical-application L-PBF parts relies heavily on expensive non-destructive inspection techniques, which significantly limits the use-cases of L-PBF. In situ monitoring of the process promises a less expensive alternative to ex situ testing, but existing sensor technologies and data analysis techniques struggle to detect sub-surface flaws (e.g., porosity and cracking) on production-scale L-PBF printers. In this work, an in situ NDE (INDE) system was engineered to detect subsurface flaws detected in X-Ray Computed Tomography (XCT) directly from process monitoring data. A multilayer, multimodal data input allowed the INDE system to detect numerous subsurface flaws in the size range of 200–1000µm using a novel human-in-the-loop annotation procedure. Furthermore, a framework was established for generating probability-of-detection (POD) and probability-of-false-alarm (PFA) curves compliant with NDE standards by systematically comparing instances of detected subsurface flaws to post-build XCT data. Here, we also introduce for the first time in the AM in situ sensing literature the a 90/95 – the flaw size corresponding to a 90% detection rate on the lower 95% confidence interval of the POD curve. The INDE system successfully demonstrated POD capabilities commensurate with traditional NDE methods. Traditional ML performance metrics were also shown to be inadequate for assessing the ability of the INDE system’s flaw detection performance. It is the hope of the authors that future studies will adopt the POD and PFA approach outlined here to provide better insight into the utility of process monitoring for AM.

36 MATERIALS SCIENCE↗

DOE OSTI2020

Analysis and improvement of the hot disk transient plane source method for low thermal conductivity materials

The hot disk transient plane source (TPS) method is a widely used standard technique (ISO 22007-2) for the characterization of thermal properties of materials, especially the thermal conductivity, k. Despite its well-established reliability for a wide variety of common materials, the hot disk TPS method is also known to suffer from a substantial systematic errors when applied to low-k thermal insulation materials. Here, we present a combined numerical and experimental study on the influence of the geometry of hot disk sensor on measured value of low-k materials. We demonstrate that the error is strongly affected by the finite thickness and thermal mass of the sensor's insulation layer was well as the corresponding increase of the effective heater size beyond the radius of the embedded metal heater itself. We also numerically investigate the dependence of the error on the sample thermal properties, confirming that the errors are worse in low-k samples. A simple correction function is also provided, which converts the apparent (erroneous) result from a standard hot disk TPS measurement to a more accurate value. A standard polyimide sensor was also optimized using both wet and dry etching to provide more accurate measurement directly. Experimentally corrected value of k for Airloy x56 aerogel and a commercial silica aerogel using the numerical correction factor derived based on the standard TPS sensor is in excellent agreement with the directly measured value from the TPS sensor using the optimized polyimide sensor. Finally, both of these methods can reduce the errors to less than 4% as compared to around 40% error of overestimation from raw values measured with the pristine sensor. Such results show that both the numerical correction to a pristine senor or an optimized sensor are capable of providing highly accurate value of thermal conductivity for such materials.

36 MATERIALS SCIENCE↗

DOE OSTI2022

Neural-network learning of SPOD latent dynamics

Here, we aim to reconstruct the latent space dynamics of high dimensional, quasi-stationary systems using model order reduction via the spectral proper orthogonal decomposition (SPOD). The proposed method is based on three fundamental steps: in the first, once that the mean flow field has been subtracted from the realizations (also referred to as snapshots), we compress the data from a high-dimensional representation to a lower dimensional one by constructing the SPOD latent space; in the second, we build the time-dependent coefficients by projecting the snapshots containing the fluctuations onto the SPOD basis and we learn their evolution in time with the aid of recurrent neural networks; in the third, we reconstruct the high-dimensional data from the learnt lower -dimensional representation. The proposed method is demonstrated on two different test cases, namely, a compressible jet flow, and a geophysical problem known as the Madden-Julian Oscillation. An extensive comparison between SPOD and the equivalent POD-based counterpart is provided and differences between the two approaches are highlighted. The numerical results suggest that the proposed model is able to provide low rank predictions of complex statistically stationary data and to provide insights into the evolution of phenomena characterized by specific range of frequencies. The comparison between POD and SPOD surrogate strategies highlights the need for further work on the characterization of the interplay of error between data reduction techniques and neural network forecasts.

97 MATHEMATICS AND COMPUTING↗