Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “model evaluation”

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 991 records · Page 55

A Model-Model and Data-Model Comparison for the Early Eocene Hydrological Cycle

A range of proxy observations have recently provided constraints on how Earth's hydrological cycle responded to early Eocene climatic changes. However, comparisons of proxy data to general circulation model (GCM) simulated hydrology are limited and inter-model variability remains poorly characterised. In this work, we undertake an intercomparison of GCM-derived precipitation and P - E distributions within the extended EoMIP ensemble (Eocene Modelling Intercomparison Project; Lunt et al., 2012), which includes previously published early Eocene simulations performed using five GCMs differing in boundary conditions, model structure, and precipitation-relevant parameterisation schemes. We show that an intensified hydrological cycle, manifested in enhanced global precipitation and evaporation rates, is simulated for all Eocene simulations relative to the preindustrial conditions. This is primarily due to elevated atmospheric paleo-CO2, resulting in elevated temperatures, although the effects of differences in paleogeography and ice sheets are also important in some models. For a given CO2 level, globally averaged precipitation rates vary widely between models, largely arising from different simulated surface air temperatures. Models with a similar global sensitivity of precipitation rate to temperature (dP=dT ) display different regional precipitation responses for a given temperature change. Regions that are particularly sensitive to model choice include the South Pacific, tropical Africa, and the Peri-Tethys, which may represent targets for future proxy acquisition. A comparison of early and middle Eocene leaf-fossil-derived precipitation estimates with the GCM output illustrates that GCMs generally underestimate precipitation rates at high latitudes, although a possible seasonal bias of the proxies cannot be excluded. Models which warm these regions, either via elevated CO2 or by varying poorly constrained model parameter values, are most successful in simulating a match with geologic data. Further data from low-latitude regions and better constraints on early Eocene CO2 are now required to discriminate between these model simulations given the large error bars on paleoprecipitation estimates. Given the clear differences between simulated precipitation distributions within the ensemble, our results suggest that paleohydrological data offer an independent means by which to evaluate model skill for warm climates.

Boundary conditions↗

A fast reduced model for a shell-and-tube based latent heat thermal energy storage heat exchanger and its application for cost optimal design by nonlinear programming

Numerical simulation of latent heat thermal energy storage (LHTES) systems plays a fundamental role in studying the physical process and guiding the engineering design. Discretization of the PDEs describing the nonlinear solidification/melting process of phase change materials (PCMs) leads to a large-scale complex dynamics system, where the system behavior depends on a set of parameters. In a design setting, repeated model evaluations are required over the set of parameters results in significant computational burden. In this paper, an explicit analytic solution was built for the propagation of the solidification front in a cylindrical coordinate. The analytic solution approach is further employed to develop a low computational reduced model (RM) as a module for a shell-and-tube based LHTES heat exchanger. The levelized Cost of Energy (LCOE) is used as a design metric and the RM model is used to apply system-level constraints in the nonlinear programming formulation that facilitates efficient global optimal design of the PCM properties, flow conditions and tube geometries. The use of LCOE as the design metric prevents over design of the heat transfer rate and also establishes a fair ground for evaluation of different thermal storage technologies and their integrated applications with other systems. Optimal results showed that a higher effectiveness results in a higher LCOE; the velocity of the HTF and the length of the channel are highly correlated with each other; both larger PCM latent energy and conductivity result in lower LCOE.

25 ENERGY STORAGE↗

Modeling Organic Aerosols during MILAGRO: Application of the CHIMERE Model and Importance of Biogenic Secondary Organic Aerosols

The meso-scale chemistry-transport model CHIMERE is used to assess our understanding of major sources and formation processes leading to a fairly large amount of organic aerosols [OA, including primary OA (POA) and secondary OA (SOA)] observed in Mexico City during the MILAGRO field project (March 2006). Chemical analyses of submicron aerosols from aerosol mass spectrometers (AMS) indicate that organic particles found in the Mexico City basin have a large fraction of oxygenated organic species (OOA), which have strong correspondence with SOA, and that their production actively continues downwind of the city. The SOA formation is modeled here by the first-generation oxidation of anthropogenic (i.e., aromatics, alkanes) and biogenic (i.e., monoterpenes and isoprene) precursors and their partitioning into both organic and aqueous phases. The near-surface model evaluation shows that predicted OA correlates reasonably well with measurements during the campaign, however it remains a factor of 2 lower than the measured total OA. Fairly good agreement is found between predicted and observed POA within the city suggesting that anthropogenic and biomass burning emissions are reasonably captured. Consistent with previous studies in Mexico City, large discrepancies are encountered for SOA species, with a factor of 5-10 model underestimate. When only anthropogenic SOA precursors were considered, the model was able to reproduce within a factor of two the sharp increase in SOA concentrations during the late morning at both urban and near-urban locations. However, predicted SOA concentrations were unrealistically low when photochemistry was not active, especially overnight. These nighttime discrepancies were not significantly reduced when greatly enhanced partitioning to the aerosol phase was assumed. Model sensitivity results suggest that observed nighttime SOA concentrations are strongly influenced by the regional background (~2µg/m3) from biogenic origin, which is transported from the coastal regions into the Mexico City basin. The relative contribution of biogenic SOA to monthly mean SOA levels is estimated to be more than 30% within the city and up to 65-90% at the regional scale (even in the immediate vicinity of the city), which is consistent with measurements of modern carbon during low biomass burning periods. The anthropogenic emissions of isoprene and its nighttime oxidation by NO3 were also found to enhance the SOA mean concentrations within the city by an additional 15%. Our results confirm the large underestimation of the SOA production by traditional models in polluted regions (estimated to 10-20 Tons within the Mexico City metropolitan area during the daytime), and emphasize for the first time the role of biogenic precursors in this region, indicating that they cannot be neglected in modeling studies.

09 BIOMASS FUELS↗

Evaluation of HEU-Beryllium Benchmark Experiments to Improve Computational Analysis of Space Reactors

An assessment was previously performed to evaluate modeling capabilities and quantify preliminary biases and uncertainties associated with the modeling methods and data utilized in designing a nuclear reactor such as a beryllium-reflected, highly-enriched-uranium (HEU)-O2 fission surface power (FSP) system for space nuclear power. The conclusion of the previous study was that current capabilities could preclude the necessity of a cold critical test of the FSP; however, additional testing would reduce uncertainties in the beryllium and uranium cross-section data and the overall uncertainty in the computational models. A series of critical experiments using HEU metal were performed in the 1960s and 1970s in support of criticality safety operations at the Y-12 Plant. Of the hundreds of experiments, three were identified as fast-fission configurations reflected by beryllium metal. These experiments have been evaluated as benchmarks for inclusion in the International Handbook of Evaluated Criticality Safety Benchmark Experiments (IHECSBE). Further evaluation of the benchmark experiments was performed using the sensitivity and uncertainty analysis capabilities of SCALE 6. The data adjustment methods of SCALE 6 have been employed in the validation of an example FSP design model to reduce the uncertainty due to the beryllium cross section data.

Bess, John↗

An Evaluation of Cosmological Models from the Expansion and Growth of Structure Measurements

Here, we compare a large suite of theoretical cosmological models to observational data from the cosmic microwave background, baryon acoustic oscillation measurements of expansion, Type Ia supernova measurements of expansion, redshift space distortion measurements of the growth of structure, and the local Hubble constant. Our theoretical models include parametrizations of dark energy as well as physical models of dark energy and modified gravity. We determine the constraints on the model parameters, incorporating the redshift space distortion data directly in the analysis. To determine whether models can be ruled out, we evaluate the p-value (the probability under the model of obtaining data as bad or worse than the observed data). In our comparison, we find the well-known tension of H 0 with the other data; no model resolves this tension successfully. Among the models we consider, the large-scale growth of structure data does not affect the modified gravity models as a category particularly differently from dark energy models; it matters for some modified gravity models but not others, and the same is true for dark energy models. We compute predicted observables for each model under current observational constraints, and identify models for which future observational constraints will be particularly informative.

79 ASTRONOMY AND ASTROPHYSICS↗

RACORO aerosol data processing

The RACORO aerosol data (cloud condensation nuclei (CCN), condensation nuclei (CN) and aerosol size distributions) need further processing to be useful for model evaluation (e.g., GCM droplet nucleation parameterizations) and other investigations. These tasks include: (1) Identification and flagging of 'splash' contaminated Twin Otter aerosol data. (2) Calculation of actual supersaturation (SS) values in the two CCN columns flown on the Twin Otter. (3) Interpolation of CCN spectra from SGP and Twin Otter to 0.2% SS. (4) Process data for spatial variability studies. (5) Provide calculated light scattering from measured aerosol size distributions. Below we first briefly describe the measurements and then describe the results of several data processing tasks that which have been completed, paving the way for the scientific analyses for which the campaign was designed. The end result of this research will be several aerosol data sets which can be used to achieve some of the goals of the RACORO mission including the enhanced understanding of cloud-aerosol interactions and improved cloud simulations in climate models.

54 ENVIRONMENTAL SCIENCES↗

Π4U: A high performance computing framework for Bayesian uncertainty quantification of complex models

We present Π4U,{sup 1} an extensible framework, for non-intrusive Bayesian Uncertainty Quantification and Propagation (UQ+P) of complex and computationally demanding physical models, that can exploit massively parallel computer architectures. The framework incorporates Laplace asymptotic approximations as well as stochastic algorithms, along with distributed numerical differentiation and task-based parallelism for heterogeneous clusters. Sampling is based on the Transitional Markov Chain Monte Carlo (TMCMC) algorithm and its variants. The optimization tasks associated with the asymptotic approximations are treated via the Covariance Matrix Adaptation Evolution Strategy (CMA-ES). A modified subset simulation method is used for posterior reliability measurements of rare events. The framework accommodates scheduling of multiple physical model evaluations based on an adaptive load balancing library and shows excellent scalability. In addition to the software framework, we also provide guidelines as to the applicability and efficiency of Bayesian tools when applied to computationally demanding physical models. Theoretical and computational developments are demonstrated with applications drawn from molecular dynamics, structural dynamics and granular flow.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

A decentralized linear quadratic control design method for flexible structures

A decentralized suboptimal linear quadratic control design procedure which combines substructural synthesis, model reduction, decentralized control design, subcontroller synthesis, and controller reduction is proposed for the design of reduced-order controllers for flexible structures. The procedure starts with a definition of the continuum structure to be controlled. An evaluation model of finite dimension is obtained by the finite element method. Then, the finite element model is decomposed into several substructures by using a natural decomposition called substructuring decomposition. Each substructure, at this point, still has too large a dimension and must be reduced to a size that is Riccati-solvable. Model reduction of each substructure can be performed by using any existing model reduction method, e.g., modal truncation, balanced reduction, Krylov model reduction, or mixed-mode method. Then, based on the reduced substructure model, a subcontroller is designed by an LQ optimal control method for each substructure independently. After all subcontrollers are designed, a controller synthesis method called substructural controller synthesis is employed to synthesize all subcontrollers into a global controller. The assembling scheme used is the same as that employed for the structure matrices. Finally, a controller reduction scheme, called the equivalent impulse response energy controller (EIREC) reduction algorithm, is used to reduce the global controller to a reasonable size for implementation. The EIREC reduced controller preserves the impulse response energy of the full-order controller and has the property of matching low-frequency moments and low-frequency power moments. An advantage of the substructural controller synthesis method is that it relieves the computational burden associated with dimensionality. Besides that, the SCS design scheme is also a highly adaptable controller synthesis method for structures with varying configuration, or varying mass and stiffness properties.

Su, Tzu-Jeng↗

Integrated Model Reduction and Control of Aircraft with Flexible Wings

This paper presents an integrated approach to the modeling and control of aircraft with exible wings. The coupled aircraft rigid body dynamics with a high-order elastic wing model can be represented in a nite dimensional state-space form. Given a set of desired output covariance, a model reduction process is performed by using the weighted Modal Cost Analysis (MCA). A dynamic output feedback controller, which is designed based on the reduced-order model, is developed by utilizing output covariance constraint (OCC) algorithm, and the resulting OCC design weighting matrix is used for the next iteration of the weighted cost analysis. This controller is then validated for full-order evaluation model to ensure that the aircraft's handling qualities are met and the uttering motion of the wings suppressed. An iterative algorithm is developed in CONDUIT environment to realize the integration of model reduction and controller design. The proposed integrated approach is applied to NASA Generic Transport Model (GTM) for demonstration.

AEROELASTIC↗

Retrieving the global distribution of the threshold of wind erosion from satellite data and implementing it into the Geophysical Fluid Dynamics Laboratory land–atmosphere model (GFDL AM4.0/LM4.0)

Dust emission is initiated when surface wind velocities exceed the threshold of wind erosion. Many dust models used constant threshold values globally. Here we use satellite products to characterize the frequency of dust events and land surface properties. By matching this frequency derived from Moderate Resolution Imaging Spectroradiometer (MODIS) Deep Blue aerosol products with surface winds, we are able to retrieve a climatological monthly global distribution of the wind erosion threshold (V(threshold)) over dry and sparsely vegetated surfaces. This monthly two-dimensional threshold velocity is then implemented into the Geophysical Fluid Dynamics Laboratory coupled land–atmosphere model (AM4.0/LM4.0). It is found that the climatology of dust optical depth (DOD) and total aerosol optical depth, surface PM10 dust concentrations, and the seasonal cycle of DOD are better captured over the “dust belt” (i.e., northern Africa and the Middle East) by simulations with the new wind erosion threshold than those using the default globally constant threshold. The most significant improvement is the frequency distribution of dust events, which is generally ignored in model evaluation. By using monthly rather than annual mean V(threshold), all comparisons with observations are further improved. The monthly global threshold of wind erosion can be retrieved under different spatial resolutions to match the resolution of dust models and thus can help improve the simulations of dust climatology and seasonal cycles as well as dust forecasting.

Bing Pu↗

Statistical uncertainty of eddy covariance CO 2 fluxes inferred using a residual bootstrap approach

High-frequency eddy-covariance measurements of net ecosystem CO 2 exchange (NEE) with the atmosphere are valuable resources for model parameterization, calibration, and validation. However, uncertainties in measured data, i.e., data gaps and inherent random errors, create problems for researchers attempting to quantify uncertainties in model projections of terrestrial ecosystem carbon cycling. Here, we demonstrate that a model-data fusion method (residual bootstrap) produces defensible annual NEE sums, through mimicking the behavior of random errors, filling missing values, and simulating gap-filling biases. This study estimated annual NEE sums for 53 site-years based on nine eddy-covariance tower sites in the USA, and found that our annual estimates were, in most cases, comparable in magnitude with those obtained from AmeriFlux gap-filled data. Additionally, compared to the AmeriFlux standardized gap-filling, our approach provides better NEE estimates for moderate to longer, and more frequent, data gaps. Annual accumulated uncertainties in NEE at the 95% confidence level were ±30gCm -2 year -1 for evergreen needleleaf forests; ±60gCm -2 year -1 for deciduous broadleaf forests; and ±80gCm -2 year -1 for croplands. The residual bootstrap performed worst when gap length was greater than one month or data exclusion greater than 90% during the growing season, common to other gap-filling techniques. However, this study produced robust results for most site years when monthly data coverage during the growing season is not extremely low. We therefore suggest that the inclusion of NEE uncertainty estimates and better estimation for moderate to longer, and more frequent, data gaps as provided by the residual bootstrap approach can be beneficial for ecosystem model evaluation.

54 ENVIRONMENTAL SCIENCES↗

Probabilistic Methods for Cyclical and Coupled Systems with Changing Failure Rates

Advancements in nuclear system designs with automated control features provide many benefits, but can lead to complex coupled systems and dynamic failure scenarios. This is especially true for microreactor designs where components are not expected to be replaced during the reactor’s lifetime. Hence, the life of the system, in addition to the safety, needs to be evaluated. Modeling these sequences of time-dependent events requires addressing cyclical processes and changing failure rates in ways that represent the actual system dynamics in contrast to a single sampling for a component’s time to failure. This research presents two distinct analytical methods for several failure distributions that evaluate a final time to failure used for different scenarios where the time to failure must be sampled multiple times. The first method is used when evaluating a component whose failure rate increases due to an outside event after the initial sampling but before the initially sampled time to failure. The second method is used when evaluating multiple identical components or a component that has been replaced with a new identical version before the second sampling. The two methods were implemented in a few representative case studies developed in the dynamic probabilistic risk assessment tool Event Modeling Risk Assessment using Linked Diagrams. Overall, this paper provides guidelines on how these approaches give a more realistic and accurate dynamic probabilistic risk assessment of complex systems.

97 MATHEMATICS AND COMPUTING↗

Evaluation of reliability modeling tools for advanced fault tolerant systems

The Computer Aided Reliability Estimation (CARE III) and Automated Reliability Interactice Estimation System (ARIES 82) reliability tools for application to advanced fault tolerance aerospace systems were evaluated. To determine reliability modeling requirements, the evaluation focused on the Draper Laboratories' Advanced Information Processing System (AIPS) architecture as an example architecture for fault tolerance aerospace systems. Advantages and limitations were identified for each reliability evaluation tool. The CARE III program was designed primarily for analyzing ultrareliable flight control systems. The ARIES 82 program's primary use was to support university research and teaching. Both CARE III and ARIES 82 were not suited for determining the reliability of complex nodal networks of the type used to interconnect processing sites in the AIPS architecture. It was concluded that ARIES was not suitable for modeling advanced fault tolerant systems. It was further concluded that subject to some limitations (the difficulty in modeling systems with unpowered spare modules, systems where equipment maintenance must be considered, systems where failure depends on the sequence in which faults occurred, and systems where multiple faults greater than a double near coincident faults must be considered), CARE III is best suited for evaluating the reliability of advanced tolerant systems for air transport.

Baker, Robert↗

Evaluating indoor exposure modeling alternatives for LCA: A case study in the vehicle repair industry

We evaluated three exposure models with data obtained from measurements among workers who use"aerosol" solvent products in the vehicle repair industry and with field experiments using these products to simulate the same exposure conditions. The three exposure models were the: 1) homogeneously-mixed-one-box model, 2) multi-zone model, and 3) eddy-diffusion model. Temporally differentiated real-time breathing zone volatile organic compound (VOC) concentration measurements, integrated far-field area samples, and simulated experiments were used in estimating parameters, such as emission rates, diffusivity, and near-field dimensions. We assessed differences in model input requirements and their efficacy for predictive modeling. The One-box model was not able to resemble the temporal profile of exposure concentrations, but it performed well concerning time-weighted exposure over extended time periods. However, this model required an adjustment for spatial concentration gradients. Multi-zone models and diffusion-models may solve this problem. However, we found that the reliable use of both these models requires extensive field data to appropriately define pivotal parameters such as diffusivity or near-field dimensions. We conclude that it is difficult to apply these models for predicting VOC exposures in the workplace. However, for comparative exposure scenarios in life-cycle assessment they may be useful.

54 ENVIRONMENTAL SCIENCES↗

Validation of FSP Reactor Design with Sensitivity Studies of Beryllium-Reflected Critical Assemblies

The baseline design for space nuclear power is a fission surface power (FSP) system: sodium-potassium (NaK) cooled, fast spectrum reactor with highly-enriched-uranium (HEU)-O 2 fuel, stainless steel (SS) cladding, and beryllium reflectors with B4C control drums. Previous studies were performed to evaluate modeling capabilities and quantify uncertainties and biases associated with analysis methods and nuclear data. Comparison of Zero Power Plutonium Reactor (ZPPR)-20 benchmark experiments with the FSP design indicated that further reduction of the total design model uncertainty requires the reduction in uncertainties pertaining to beryllium and uranium cross-section data. Further comparison with three beryllium-reflected HEU-metal benchmark experiments performed at the Oak Ridge Critical Experiments Facility (ORCEF) concluded the requirement that experimental validation data have similar cross section sensitivities to those found in the FSP design. A series of critical experiments was performed at ORCEF in the 1960s to support the Medium Power Reactor Experiment (MPRE) space reactor design. The small, compact critical assembly (SCCA) experiments were graphite- or beryllium-reflected assemblies of SS-clad, HEU-O 2 fuel on a vertical lift machine. All five configurations were evaluated as benchmarks. Two of the five configurations were beryllium reflected, and further evaluated using the sensitivity and uncertainty analysis capabilities of SCALE 6.1. Validation of the example FSP design model was successful in reducing the primary uncertainty constituent, the Be(n,n) reaction, from 0.28 %$\delta$k/k to 0.0004 %$\delta$k/k. Further assessment of additional reactor physics measurements performed on the SCCA experiments may serve to further validate FSP design and operation.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗

SU-E-T-128: Dosimetric Evaluation of MLC Modeling in Pinnacle V9.2 for Varian TrueBeam STx

Purpose: To evaluate the effects of MLC modeling after commissioning the Varian TrueBeam LINAC in Pinnacle version 9.2. Methods: Stepand-shoot IMRT QAs were investigated when we observed our measured absolute dose results using ion chamber (Capintec PR-05P) were uncharacteristically low; about 4–5% compared to doses calculated by Pinnacle{sup 3} (Phillips, Madison, WI). This problem was predominant for large and highly modulated head and neck (HN) treatments. Intuitively we knew this had to be related to shortcomings in the MLC modeling in Pinnacle. Using film QA we were able to iteratively adjust the MLC parameters. We confirmed results by re-testing five failed IMRT QA patients; and ion chamber measurements were verified in Quasar anthropomorphic phantom. Results: After commissioning the LINAC in Pinnacle version 9.2, the MLC transmission for 6X, 10X and 15X were 2.0%, 1.7% and 2.0%, respectively, and additional Interleaf leakage for all three energies was 0.5%. These parameters were obtained from profiles scanned with an Edge detector (Sun Nuclear, Melbourne, FL) during machine commissioning. A Verification testing with radiographic EDR2 film (Kodak, Rochester, NY) measurement was performed by creating a closed MLC leaf pattern and analyzing using RIT software (RIT, Colorado Springs, CO). This reduced MLC transmission for 6X, 10X and 15X to 0.7%, 0.9% and 0.9%, respectively; while increasing additional Interleaf leakage for all three energies to 1.0%. Conclusion: Radiographic film measurements were used to correct MLC transmission values for step and shoot IMRT fields used in Pinnacle version 9.2. After adjusting the MLC parameters to correlate with the film QA, there was still very good agreement between the Pinnacle model and commissioning data. Using the same QA methodology, we were also able to improve the beam models for the Varian C-series linacs, Novalis-Tx, and TrueBeam M-120 linacs.

07 ISOTOPE AND RADIATION SOURCES↗

Ozone Climatology

The primary role of models in the assessment process is to predict changes to ozone. It is crucial therefore that the ability of the models to reproduce the actual distribution of ozone be tested. Historically, maps of the ozone column (latitude by month) have been used for this purpose. In MM I a climatology was developed for the vertical distribution of ozone for 15-60 km, based on SBUV data for 1979-80. SBUV profiles are reported with vertical resolution of approx. 5 km, but the true resolution is lower, approx. 8 km above the ozone maximum and approx. 15 km for 10-25 km. The climatology was considered valid to about 20-30% at 20 km and to 50% at 15 km. Comparisons were made with models in mixing ratio (ppm), which emphasizes the middle and upper stratosphere. A new ozone climatology was developed for the vertical distribution of ozone for MM II. Our goal was to develop a product that could be used to evaluate models in the lower stratosphere, the region where most of the ozone column resides and where most of the ozone loss is occurring, as well as the middle and upper stratosphere.

Logan, Jennifer A.↗

A Recursive Multi-step Machine Learning Approach for Airport Configuration Prediction

Airport configuration selection is a complex decision-making process that involves several operational and human factors. In this paper we propose a novel recursive multi-step machine learning (ML) approach to predict airport configuration. The multi-step approach guarantees stability of the predicted configuration by taking as input the configuration predicted at the previous time step. The features of the proposed model include weather data, future arrival and departure counts and current configuration. Due to the importance of arrival and departure counts in predicting the airport configuration, arrival counts are calculated using landing time predictions selected from physics-based landing time predictions available in FAA System Wide Information Management data feeds for each flight. The selection rules were developed and refined to select the most accurate time for different phases of flight. The proposed model predicts the airport configurations up to 6 hours ahead. In this paper we show the predictive performance of the proposed model for six major US airports, including Charlotte Douglas International Airport (CLT), Dallas/Fort Worth International Airport (DFW), John F. Kennedy International Airport (JFK), Newark Liberty International Airport (EWR), LaGuardia Airport (LGA) and Dallas Love Field Airport (DAL). We trained and evaluated models on 2019 and 2020 data in order to study the effect of the pandemic and how changes in traffic patterns affected the performance of the proposed model. Results are compared with a baseline assuming no airport configuration changes. In our results for DFW, we obtained a prediction accuracy of 89.3% for 3 hours ahead prediction, and 82.8% for 6 hours ahead when applied on 2019 data.

machine learning↗