Search NASA⌕ Search

SEARCH · Search NASA

Results for “explicit model predictive control”

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 37 records · Page 2

Application of optimization techniques to the design of a flutter suppression control law for the DAST ARW-2

The design of a candidate flutter suppression (FS) control law for the symmetric degrees of freedom for the DAST ARW-2 aircraft is discussed. The results illustrate the application of several currently employed control law design techniques. Subsequent designs, obtained as the mathematical model of the ARW-2 is updated, are expected to employ similar methods and to provide a control law whose performance will be flight tested. This study represents one of the steps necessary to provide an assessment of the validity of applying current control law synthesis and analysis techniques in the design of actively controlled aircraft. Mathematical models employed in the control law design and evaluation phases are described. The control problem is specified by presenting the flutter boundary predicted for the uncontrolled aircraft and by defining objectives and constraints that the controller should satisfy. A full-order controller is obtained by using Linear Quadratic Gaussian (LQG) techniques. The process of obtaining an implementable reduced-order controller is described. One example is also shown in which constrained optimization techniques are utilized to explicitly include robustness criteria within the design algorithm.

Adams, W. M., Jr.↗

The effect of gravity induced free convection upon the melting phenomena of a finite paraffin slab for thermal control

Two theoretical models were developed to predict the thermal response of the phase change material to a given hot plate temperature. A two-dimensional pure conduction model was developed to predict the melting of the phase change material when heat transfer was a function of conduction. A combined conduction-convection model, also two-dimensional, was developed to predict the phase change phenomena when heat transfer was a function of conduction and gravity-induced free convection. Both models were solved using explicit finite difference approximations on a digital computer. The experimental equipment consisted of a rectangular cell utilizing a heat chamber, an expansion chamber, and a test chamber; a sixteen channel multipoint recorder, and a fluid flow system. The recorder monitered hot and cold plate temperatures and interior node temperatures at two second intervals. A comparison of theoretical temperature profiles and experimental temperature profiles is presented for six runs at various angles of inclination of the test cell with respect to the horizontal direction. A detailed discussion of results is presented.

Bain, R. L.↗

Cirrus Parcel Model Comparison Project: The Critical Components to Simulate Cirrus Initiation Explicitly - Phase 1

The Cirrus Parcel Model Comparison Project, a project of the GCSS (GEWEX Cloud System Studies) Working Group on Cirrus Cloud Systems, involves the systematic comparison of current models of ice crystal nucleation and growth for specified, typical, cirrus cloud environments. In Phase I of the project reported here, simulated cirrus cloud microphysical properties are compared for situations of "warm" (40 C) and "cold" (-60 C) cirrus, both subject to updrafts of 4, 20 and 100 centimeters per second. Five models participated. The various models employ explicit microphysical schemes wherein the size distribution of each class of particles (aerosols and ice crystals) is resolved into bins or treated separately. Simulations are made including both the homogeneous and heterogeneous ice nucleation mechanisms. A single initial aerosol population of sulfuric acid particles is prescribed for all simulations. To isolate the treatment of the homogeneous freezing (of haze droplets) nucleation process, the heterogeneous nucleation mechanism is disabled for a second parallel set of simulations. Qualitative agreement is found for the homogeneous-nucleation- only simulations, e.g., the number density of nucleated ice crystals increases with the strength of the prescribed updraft. However, significant quantitative differences are found. Detailed analysis reveals that the homogeneous nucleation rate, haze particle solution concentration, and water vapor uptake rate by ice crystal growth (particularly as controlled by the deposition coefficient) are critical components that lead to differences in predicted microphysics. Systematic bias exists between results based on a modified classical theory approach and models using an effective freezing temperature approach to the treatment of nucleation. Each approach is constrained by critical freezing data from laboratory studies, but each includes assumptions that can only be justified by further laboratory research. Consequently, it is not yet clear if the two approaches can be made consistent. Large haze particles may deviate considerably from equilibrium size in moderate to strong updrafts (20-100 centimeters per second) at -60 C when the commonly invoked equilibrium assumption is lifted. The resulting difference in particle-size- dependent solution concentration of haze particles may significantly affect the ice particle formation rate during the initial nucleation interval. The uptake rate for water vapor excess by ice crystals is another key component regulating the total number of nucleated ice crystals. This rate, the product of particle number concentration and ice crystal diffusional growth rate, which is particularly sensitive to the deposition coefficient when ice particles are small, modulates the peak particle formation rate achieved in an air parcel and the duration of the active nucleation time period. The effects of heterogeneous nucleation are most pronounced in weak updraft situations. Vapor competition by the heterogeneously nucleated ice crystals may limit the achieved ice supersaturation and thus suppresses the contribution of homogeneous nucleation. Correspondingly, ice crystal number density is markedly reduced. Definitive laboratory and atmospheric benchmark data are needed for the heterogeneous nucleation process. Inter-model differences are correspondingly greater than in the case of the homogeneous nucleation process acting alone.

Lin, Ruei-Fong↗

A comparison of motor submodels in the optimal control model

Properties of several structural variations in the neuromotor interface portion of the optimal control model (OCM) are investigated. For example, it is known that commanding control-rate introduces an open-loop pole at S=O and will generate low frequency phase and magnitude characteristics similar to experimental data. However, this gives rise to unusually high sensitivities with respect to motor and sensor noise-ratios, thereby reducing the models' predictive capabilities. Relationships for different motor submodels are discussed to show sources of these sensitivities. The models investigated include both pseudo motor-noise and actual (system driving) motor-noise characterizations. The effects of explicit proprioceptive feedback in the OCM is also examined. To show graphically the effects of each submodel on system outputs, sensitivity studies are included, and compared to data obtained from other tests.

Lancraft, R. E.↗

B33C-0612: Evaluation of Simulated Biospheric Carbon Dioxide Fluxes and Atmospheric Concentrations Using Global in Situ Observations

Atmospheric mixing ratios of carbon dioxide (CO2) are largely controlled by anthropogenic emission sources and biospheric sources/sinks. Global biospheric fluxes of CO2 are controlled by complex processes facilitating the exchange of carbon between terrestrial ecosystems and the atmosphere. These processes which play a key role in these terrestrial ecosystem-atmosphere carbon exchanges are currently not fully understood, resulting in large uncertainties in the quantification of biospheric CO2 fluxes. Current models with these inherent deficiencies have difficulties simulating the global carbon cycle with high accuracy. We are developing a new modeling platform, GEOS-Chem-CASA by integrating the year-specific NASA-CASA (National Aeronautics and Space Administration - Carnegie Ames Stanford Approach) biosphere model with the GEOS-Chem (Goddard Earth Observation System-Chemistry) chemical transport model to improve the simulation of atmosphere-terrestrial ecosystem carbon exchange. We use NASA-CASA to explicitly represent the exchange of CO2 between terrestrial ecosystem and atmosphere by replacing the baseline GEOS-Chem land net CO2 flux and forest biomass burning CO2 emissions. We will present the estimation and evaluation of these "bottom-up" land CO2 fluxes, simulated atmospheric mixing ratios, and forest disturbance changes over the last decade. In addition, we will present our initial comparison of atmospheric column-mean dry air mole fraction of CO2 predicted by the model and those retrieved from NASA's OCO-2 (Orbiting Carbon Observatory-2) satellite instrument and model-predicted surface CO2 mixing ratios with global in situ observations. This evaluation is the first step necessary for our future work planned to constrain the estimates of biospheric carbon fluxes through "top-down" inverse modeling, which will improve our understanding of the processes controlling atmosphere-terrestrial ecosystem greenhouse gas exchanges, especially over regions which lack in situ observations.

in situ observations↗

Effectiveness of basic display augmentation in vehicular control by visual field cues

The paper investigates the effectiveness of different basic display augmentation concepts - fixed reticle, velocity vector, and predicted future vehicle path - for RPVs controlled by a vehicle-mounted TV camera. The task is lateral manual control of a low flying RPV along a straight reference line in the presence of random side gusts. The man-machine system and the visual interface are modeled as a linear time-invariant system. Minimization of a quadratic performance criterion is assumed to underlie the control strategy of a well-trained human operator. The solution for the optimal feedback matrix enables the explicit computation of the variances of lateral deviation and directional error of the vehicle and of the control force that are used as performance measures.

Grunwald, A. J.↗

Fault tolerant system performance modeling

With the proliferation of complex digital systems on aircraft, the need to accurately predict system performance early in the system design cycle becomes imperative. In the past, system designers have relied on ad hoc methods for evaluating performance issues. This has produced systems that have not always worked as originally intended. To alleviate these design deficiencies, formal methods, with supporting tools, must be adhered to during the system design process. The use of performance modeling tools is becoming widely accepted as a way to address timing considerations of system design. An additional incentive for the use of these tools is that they allow the system architect to analyze system component interactions (i.e., bus contention, contention of functions for a processing site, and system repair activity on application performance). This inherent flexibility can result in an explicit specification of the system architecture. This paper addresses a method that supports performance modeling of fault tolerant systems using a discrete event simulation tool. An additional focus is on lessons learned from analyzing these classes of problems. The methodology and supporting work provide system architects with the capability to specify candidate architectures and accurately predict their performance in the early stages of design, where changes to system design is most cost effective. The work has been supported under NASA contract NAS1-18099. Integrated Airframe Propulsion Control System Architecture (IAPSA II). This contract addresses methodology, analysis, and detailed design of integrated control system architectures suitable for high-performance aircraft of the 1990's.

Discrete event simulation↗

Optics Program Simplifies Analysis and Design

Engineers at Goddard Space Flight Center partnered with software experts at Mide Technology Corporation, of Medford, Massachusetts, through a Small Business Innovation Research (SBIR) contract to design the Disturbance-Optics-Controls-Structures (DOCS) Toolbox, a software suite for performing integrated modeling for multidisciplinary analysis and design. The DOCS Toolbox integrates various discipline models into a coupled process math model that can then predict system performance as a function of subsystem design parameters. The system can be optimized for performance; design parameters can be traded; parameter uncertainties can be propagated through the math model to develop error bounds on system predictions; and the model can be updated, based on component, subsystem, or system level data. The Toolbox also allows the definition of process parameters as explicit functions of the coupled model and includes a number of functions that analyze the coupled system model and provide for redesign. The product is being sold commercially by Nightsky Systems Inc., of Raleigh, North Carolina, a spinoff company that was formed by Mide specifically to market the DOCS Toolbox. Commercial applications include use by any contractors developing large space-based optical systems, including Lockheed Martin Corporation, The Boeing Company, and Northrup Grumman Corporation, as well as companies providing technical audit services, like General Dynamics Corporation

Source record↗

Very Large Eddy Simulation Technique for Noise Prediction and Control in Turbomachinery and Propulsion

The summer fellowship research project focused on further developing an advanced computational technique based on Very Large Eddy Simulation (VLES) for analysis and control of major sources of noise in turbomachinery and propulsion systems, including jet noise and fan noise. Major part of the work during the 10-week tenure dealt with implementing a low-order, implicit A-stable time-stepping scheme in the existing explicit VLES code of Dr. Ray Hixon. The preliminary plan of the work also included application of a new time marching formulation to the problem of viscous gust-airfoil interaction. Other research items selected for implementation (possibly in the future) included investigating a set of new subgrid turbulent models for the code, and code application to a number of test cases, including a supersonic jet and swirling flow downstream of a rotor stage.

Golubev, Vladimir V.↗

Transient modeling/analysis of hyperbolic heat conduction problems employing mixed implicit-explicit alpha method

This paper describes the evaluation of mixed implicit-explicit finite element formulations for hyperbolic heat conduction problems involving non-Fourier effects. In particular, mixed implicit-explicit formulations employing the alpha method proposed by Hughes et al. (1987, 1990) are described for the numerical simulation of hyperbolic heat conduction models, which involves time-dependent relaxation effects. Existing analytical approaches for modeling/analysis of such models involve complex mathematical formulations for obtaining closed-form solutions, while in certain numerical formulations the difficulties include severe oscillatory solution behavior (which often disguises the true response) in the vicinity of the thermal disturbances, which propagate with finite velocities. In view of these factors, the alpha method is evaluated to assess the control of the amount of numerical dissipation for predicting the transient propagating thermal disturbances. Numerical test models are presented, and pertinent conclusions are drawn for the mixed-time integration simulation of hyperbolic heat conduction models involving non-Fourier effects.

Tamma, Kumar K.↗

Prediction of Film Cooling on Gas Turbine Airfoils

A three-dimensional Navier-Stokes analysis tool has been developed in order to study the effect of film cooling on the flow and heat transfer characteristics of actual turbine airfoils. An existing code (Arnone et al., 1991) has been modified for the purpose. The code is an explicit, multigrid, cell-centered, finite volume code with an algebraic turbulence model. Eigenvalue scaled artificial dissipation and variable-coefficient implicit residual smoothing are used with a full-multigrid technique. Moreover, Mayle's transition criterion (Mayle, 1991) is used. The effects of film cooling have been incorporated into the code in the form of appropriate boundary conditions at the hole locations on the airfoil surface. Each hole exit is represented by several control volumes, thus providing an ability to study the effect of hole shape on the film-cooling characteristics. Comparison is fair with near mid-span experimental data for four and nine rows of cooling holes, five on the shower head, and two rows each on the pressure and suction surfaces. The computations, however, show a strong spanwise variation of the heat transfer coefficient on the airfoil surface, specially with shower-head cooling.

Garg, Vijay K.↗

CFD Modeling of the Multipurpose Hydrogen Test Bed (MHTB) Self-Pressurization and Spray Bar Mixing Experiments in Normal Gravity: Effect of the Accommodation Coefficient on the Tank Pressure

A CFD model for simulating the self-pressurization of a large scale liquid hydrogen storage tank is utilized in this paper to model the MHTB self-pressurization experiment. The kinetics-based Schrage equation is used to account for the evaporative and condensi ng interfacial mass flows in this model. The effect of the accommodation coefficient for calculating the interfacial mass transfer rate on the tank pressure during tank selfpressurization is studied. The values of the accommodation coefficient which were considered in this study vary from 1.0e-3 to 1.0e-1 for the explicit VOF model and from 1.0e-4 to 1.0e-3 for the implicit VOF model. The ullage pressure evolutions are compared against experimental data. A CFD model for controlling pressure in cryogenic storage tanks by spraying cold liquid into the ullage is also presented. The Euler-Lagrange approach is utilized for tracking the spray droplets and for modeling the interaction between the droplets and the continuous phase (ullage). The spray model is coupled with the VOF model by performing particle tracking in the ullage, removing particles from the ullage when they reach the interface, and then adding their contributions to the liquid. Droplet-ullage heat and mass transfer are modeled. The flow, temperature, and interfacial mass flux, as well as droplets trajectories, size distribution and temperatures predicted by the model are presented. The ul lage pressure and vapor temperature evolutions are compared with experimental data obtained from the MHTB spray bar mixing experiment. The effect of the accommodation coefficient for calculating the interfacial and droplet mass transfer rates on the tank pressure during mixing of the vapor using spray is studied. The values used for the accommodation coefficient at the interface vary from 1.0e-5 to 1.0e-2. The droplet accommodation coefficient values vary from 2.0e-6 to 1.0e-4.

cryogenic fluid storage↗

Atmospheric Dissolved Iron Depostiion to the Global Oceans: Effects of Oxalate-Promoted Fe Dissolution, Photochemical Redox Cycling, and Dust Mineralogy

Mineral dust deposition is suggested to be a significant atmospheric supply pathway of bioavailable iron (Fe) to Fe-depleted surface oceans. In this study, mineral dust and dissolved Fe (Fed) deposition rates are predicted for March 2009 to February 2010 using the 3-D chemical transport model GEOS-Chem implemented with a comprehensive dust-Fe dissolution scheme. The model simulates Fed production during the atmospheric transport of mineral dust taking into account inorganic and organic (oxalate)-promoted Fe dissolution processes, photochemical redox cycling between ferric (Fe(III)) and ferrous (Fe(II)) forms of Fe, dissolution of three different Fe-containing minerals (hematite, goethite, and aluminosilicates), and detailed mineralogy of windblown dust from the major desert regions. Our calculations suggest that during the yearlong simulation is approximately 0.26 Tg (1 Tg = 1012 g) of Fed was deposited to global oceanic regions. Compared to simulations only taking into account proton-promoted Fe dissolution, the addition of oxalate to the dust-Fe mobilization scheme increased total annual model-predicted Fed deposition to global oceanic regions by approximately 75%. The implementation of Fe(II)/Fe(III) photochemical redox cycling in the model allows for the distinction between different oxidation states of deposited Fed. Our calculations suggest that during the daytime, large fractions of Fed deposited to the global oceans is likely to be in Fe(II) form, while nocturnal fluxes of Fed are largely in Fe(III) form. Model simulations also show that atmospheric fluxes of Fed can be strongly influenced by the mineralogy of Fe-containing compounds. This study shows that Fed deposition to the oceans is controlled by total dust-Fe mass concentrations, mineralogy, the surface area of dust particles, atmospheric chemical composition, cloud processing, and meteorological parameters and exhibits complex and spatiotemporally variable patterns. Our study suggests that the explicit model representation of individual processes leading to Fed production within mineral dust are needed to improve the understanding of the atmospheric Fe cycle, and quantify the effect of dust-Fe on ocean biological productivity, carbon cycle, and climate.

Johnson, M. S.↗

Intercomparison of Large-Eddy Simulations of Arctic Mixed-Phase Clouds: Importance of Ice Size Distribution Assumptions

Large-eddy simulations of mixed-phase Arctic clouds by 11 different models are analyzed with the goal of improving understanding and model representation of processes controlling the evolution of these clouds. In a case based on observations from the Indirect and Semi-Direct Aerosol Campaign (ISDAC), it is found that ice number concentration, Ni, exerts significant influence on the cloud structure. Increasing Ni leads to a substantial reduction in liquid water path (LWP), in agreement with earlier studies. In contrast to previous intercomparison studies, all models here use the same ice particle properties (i.e., mass-size, mass-fall speed, and mass-capacitance relationships) and a common radiation parameterization. The constrained setup exposes the importance of ice particle size distributions (PSDs) in influencing cloud evolution. A clear separation in LWP and IWP predicted by models with bin and bulk microphysical treatments is documented and attributed primarily to the assumed shape of ice PSD used in bulk schemes. Compared to the bin schemes that explicitly predict the PSD, schemes assuming exponential ice PSD underestimate ice growth by vapor deposition and overestimate mass-weighted fall speed leading to an underprediction of IWP by a factor of two in the considered case. Sensitivity tests indicate LWP and IWP are much closer to the bin model simulations when a modified shape factor which is similar to that predicted by bin model simulation is used in bulk scheme. These results demonstrate the importance of representation of ice PSD in determining the partitioning of liquid and ice and the longevity of mixed-phase clouds.

clouds↗

Spectral Analysis of Forecast Error Investigated with an Observing System Simulation Experiment

The spectra of analysis and forecast error are examined using the observing system simulation experiment (OSSE) framework developed at the National Aeronautics and Space Administration Global Modeling and Assimilation Office (NASAGMAO). A global numerical weather prediction model, the Global Earth Observing System version 5 (GEOS-5) with Gridpoint Statistical Interpolation (GSI) data assimilation, is cycled for two months with once-daily forecasts to 336 hours to generate a control case. Verification of forecast errors using the Nature Run as truth is compared with verification of forecast errors using self-analysis; significant underestimation of forecast errors is seen using self-analysis verification for up to 48 hours. Likewise, self analysis verification significantly overestimates the error growth rates of the early forecast, as well as mischaracterizing the spatial scales at which the strongest growth occurs. The Nature Run-verified error variances exhibit a complicated progression of growth, particularly for low wave number errors. In a second experiment, cycling of the model and data assimilation over the same period is repeated, but using synthetic observations with different explicitly added observation errors having the same error variances as the control experiment, thus creating a different realization of the control. The forecast errors of the two experiments become more correlated during the early forecast period, with correlations increasing for up to 72 hours before beginning to decrease.

OSSE↗

Convergence acceleration of implicit schemes in the presence of high aspect ratio grid cells

The performance of Navier-Stokes codes are influenced by several phenomena. For example, the robustness of the code may be compromised by the lack of grid resolution, by a need for more precise initial conditions or because all or part of the flowfield lies outside the flow regime in which the algorithm converges efficiently. A primary example of the latter effect is the presence of extended low Mach number and/or low Reynolds number regions which cause convergence deterioration of time marching algorithms. Recent research into this problem by several workers including the present authors has largely negated this difficulty through the introduction of time-derivative preconditioning. In the present paper, we employ the preconditioned algorithm to address convergence difficulties arising from sensitivity to grid stretching and high aspect ratio grid cells. Strong grid stretching is particularly characteristic of turbulent flow calculations where the grid must be refined very tightly in the dimension normal to the wall, without a similar refinement in the tangential direction. High aspect ratio grid cells also arise in problems that involve high aspect ratio domains such as combustor coolant channels. In both situations, the high aspect ratio cells can lead to extreme deterioration in convergence. It is the purpose of the present paper to address the reasons for this adverse response to grid stretching and to suggest methods for enhancing convergence under such circumstances. Numerical algorithms typically possess a maximum allowable or optimum value for the time step size, expressed in non-dimensional terms as a CFL number or vonNeumann number (VNN). In the presence of high aspect ratio cells, the smallest dimension of the grid cell controls the time step size causing it to be extremely small, which in turn results in the deterioration of convergence behavior. For explicit schemes, this time step limitation cannot be exceeded without violating stability restrictions of the scheme. On the other hand, for implicit schemes, which are typically unconditionally stable, there appears to be room for improvement through careful tailoring of the time step definition based on results of linear stability analyses. In the present paper, we focus on the central-differenced alternating direction implicit (ADI) scheme. The understanding garnered from this analyses can then be applied to other implicit schemes. In order to systematically study the effects of aspect ratio and the methods of mitigating the associated problems, we use a two pronged approach. We use stability analyses as a tool for predicting numerical convergence behavior and numerical experiments on simple model problems to verify predicted trends. Based on these analyses, we determine that efficient convergence may be obtained at all aspect ratios by getting a combination of things right. Primary among these are the proper definition of the time step size, proper selection of viscous preconditioner and the precise treatment of boundary conditions. These algorithmic improvements are then applied to a variety of test cases to demonstrate uniform convergence at all aspect ratios.

Buelow, B. E. O.↗

Universal limiter for transient interpolation modeling of the advective transport equations: The ULTIMATE conservative difference scheme

A fresh approach is taken to the embarrassingly difficult problem of adequately modeling simple pure advection. An explicit conservative control-volume formation makes use of a universal limiter for transient interpolation modeling of the advective transport equations. This ULTIMATE conservative difference scheme is applied to unsteady, one-dimensional scalar pure advection at constant velocity, using three critical test profiles: an isolated sine-squared wave, a discontinuous step, and a semi-ellipse. The goal, of course, is to devise a single robust scheme which achieves sharp monotonic resolution of the step without corrupting the other profiles. The semi-ellipse is particularly challenging because of its combination of sudden and gradual changes in gradient. The ULTIMATE strategy can be applied to explicit conservation schemes of any order of accuracy. Second-order schemes are unsatisfactory, showing steepening and clipping typical of currently popular so-called high resolution shock-capturing of TVD schemes. The ULTIMATE third-order upwind scheme is highly satisfactory for most flows of practical importance. Higher order methods give predictably better step resolution, although even-order schemes generate a (monotonic) waviness in the difficult semi-ellipse simulation. Little is to be gained above ULTIMATE fifth-order upwinding which gives results close to the ultimate for which one might hope.

Leonard, B. P.↗

A Machine Learning Approach to Improve Air Traffic Management Initiatives

Collaborating closely with commercial air carriers and related organizations, the Federal Aviation Administration(FAA) regulates air traffic and ensures the safety and efficiency of air operations. Air traffic controllers make strategic decisions, such as delaying, rerouting, or canceling flights, partly based on guidance provided by the FAA’s Air TrafficControl System Command Center (ATCSCC). The guidance includes, among other things, control measures known asTraffic Management Initiatives (TMIs) designed to enhance safety and improve operational efficiency. TMIs play a crucial role in managing the demand and capacity within the U.S. National Airspace System (NAS). Two major TMIs that are routinely used (primarily to mitigate the adverse effects of bad weather) are Ground Delay Programs (GDPs) andGround Stops (GSs). In a GDP, flights destined for airports facing thunderstorm activity experience delays at their origin airports. This proactive approach minimizes the risk of routing aircraft through hazardous weather conditions and also replaces (fuel burning) airborne delays with ground delays. In a GS, a temporary restriction is imposed on the departure or arrival of aircraft at a specific airport or within a designated airspace. Although other TMIs (e.g., miles-in-trail) are also implemented as part of (air) traffic flow management in the NAS, the focus of this work is on GDPs and GSs. Since TMIs, by design, lead to flight delays or cancellations, it is crucial to put in place the right set of parameters(e.g., scope and duration of the GDP). For example, when the end time of a GDP extends beyond what is necessary, it imposes unnecessary delays on departing flights. This situation could occur as a result of inaccurate prediction of the(required) duration of the GDP based on the weather forecast. On the other hand, if a GDP ends prematurely before the underlying capacity constraints are resolved at the destination airport, it may result in airborne holding. The delicate balance lies in matching the termination of the GDP precisely with the resolution of capacity constraints, avoiding both the imposition of unnecessary ground delays and the need for airborne holding due to premature program termination.Failing to specify the right parameters for TMIs also leads to flight delays, creating a significant obstacle in managing the increasing traffic volumes causing increased work load for the controllers. To address this issue, we propose the integration of Machine Learning (ML) models in the traffic flow management(TFM) pipeline. In current operations, decisions are made by human experts based on extensive training, historical patterns, available traffic and weather data. Since we have an abundance of data from past events that tell us the likely impact of various TMIs, by ingesting historical data, properly trained ML models can offer valuable insights and aid human decision-making. With the FAA increasingly exploring advanced analytics, ML emerges as a focal point for enhancing TFM within the National Airspace System (NAS). As a first step, this study aims to provide traffic controllers with decision-making support for the issuance and adjustment of TMIs. Data analytics and machine learning have been previously employed to address some of the challenges associated with TMIs. Numerous studies have concentrated on various facets of TMI issuance, exploring factors influencing TMI parameters, including arrival rate, airport capacity, and delay prediction. For example, using weather forecasts, several statistical methods were used to produce probabilistic capacity profiles which in conjunction with deterministic models provided insights into the GDP planning process [1–4]. The downside of using deterministic models is that they rely on fixed inputs and predetermined rules, which lack the ability to account for the inherent uncertainty and variability present in real-world scenarios. In a separate series of studies, researchers aimed to predict the occurrences of GDPs and GSs. The majority of these studies utilized various supervised learning methods, including Decision Trees, Naive Bayes, Support VectorMachines, and Random Forests to analyze the influence of weather conditions and arrival demand on TMI incidents[5–8]. However, these studies primarily focused on predicting the incidence of TMIs without explicitly addressing the scope of TMIs, including their duration and their geographical coverage. Furthermore, the emphasis of these studies was largely on GDPs, given their higher frequency and longer duration when compared to GSs. A limited number of studies focused on predicting the parameters of TMIs, specifically addressing their duration and extent. In one such study focusing on optimizing the TMI parameters at San Francisco International Airport (SFO),the authors utilized a probabilistic forecast of fog [9]. They simulated various capacity scenarios based on the (fog)burn-off forecasts, selecting GDP parameters that minimized airborne and overall ground delays. However, this approach exclusively emphasizes stratus (fog) burn-off as the primary determinant of GDP and GS, neglecting other influential factors like severe weather events, runway closures, lower capacity than traffic demand, and other important variables. Given the complexity of predicting the TMI and determining its scope, we seek a more holistic approach. We aim to consider all significant factors that could impact TMIs and their parameters. What sets this research apart is the fusion of all data sources relevant to the issuance and adjustment of TMIs and it represents the first comprehensive attempt to optimize TMIs in this manner. Since this comprehensive solution involves various aspects, we break down the problem into smaller components and input all parameters into a unified model called the “TMI Adjuster”. Figure 1 shows the overall framework and the list of datasets used in each model. The objective of the TMI Adjuster module is to deliver reliable, consistent and expedited recommendations for the progression, adjustment, and termination of TMIs. The ML solution entails developing a pipeline capable of predicting the necessity of a TMI (e.g., GS or GDP) along with its various parameters. For example, in the case of a GS, this includes the scope of the GS either in terms of distance from the destination airport or based on pre-defined airspace sectors. Here, scope refers to those regions and departing airports that are subject to the GS. In this paper, we concentrate on the issuance of GSs in the three major airports in the New York area — LaGuardia(LGA), John F. Kennedy International (JFK), and Newark Liberty International (EWR). We fuse traffic, weather and other relevant aviation data from years 2017 to 2019 to train and validate the ML models. In particular, we use the following datasets: •Terminal Aerodrome Forecast (TAF): meteorological forecasts specific to each airport, issued four times a day, covering predefined time periods. •TMI data: includes all GSs and GDPs along with their respective parameters. •Aviation System Performance Metrics (ASPM): includes traffic related data such as aircraft delays, arrival, and departure rates. •Notices to Airmen (NOTAMs): utilized to extract runway closure data and manage interdependencies between terminals in close proximity. •Flight cancellation data •Airspace Flow Programs (AFP): includes information on flight airborne holdings caused by TMIs. The data preprocessing entails transforming ASPM, TMI, AFP, NOTAMs, and weather data into an hourly format and consolidating all datasets by merging them based on date and time as the primary key. The TMI Adjuster framework comprises two parallel models: one dedicated to GS and a second model focused on GDP. As previously mentioned, our specific focus is on the GS model as a multi-classification problem. In this framework, each data point of the GS model input summarizes ten hours of data. Specifically, the data loader for the GS model generates the input and output of the model as follows: at a given time step, the input includes the actual traffic, weather, and TMI data from the two-hour window before the time step, alongside the weather forecast and scheduled traffic for the next 8 hours starting from the time step. Based on this information, the output of the GS model for each time interval consists of three dimensions. The first dimension represents a binary decision on whether there should be a GS in place for the next hour or not. The second dimension is related to the scope of the GS in the United States, and the third dimension is related to the scope of the GS in Canada (i.e., to determine if the GS impacts airports in Canada).One of the challenges with TMI modeling is the sparsity of TMI events, particularly regarding its scope. To address this challenge in the scope of the GS model output, we implement grouping. The GS scope for the US region is defined based on a list of centers that should be included when the GS is in place. With 20 centers in the US, we utilized historical data to group them into 4 categories. In particular, we summarized our historical data in a graph format where nodes represent centers, and link weights are defined based on the co-occurrence of centers in the scope parameter ofTMIs. By identified strongly connected components in this graph, we were able to partition the centers into four groups. We consider two model structures for the GS Model. Firstly, a hierarchical classification model [10], where the human decision-making for a GS is of hierarchical nature. The decision-maker first decides whether there is a need fora GS, and if the answer is yes, determines the scope. A hierarchical classification model organizes the problem into a class hierarchy, typically a tree or a Directed Acyclic Graph (DAG) structure, and considers the dependency of the decision in the previous step to the next component [10]. Here, we employ the local classifier per level approach, which involves training one multi-class classifier for each level of the class hierarchy. The second structure is the independent structure. In this setting, as the name suggests, we do not consider the dependency of the decisions in the different dimensions of the output of the model. Instead, for each dimension, we train a multi-class classifier independently. Table 1 summarizes GS model statistics for training, validation and testing. The table documents the effect of limiting data to the time steps when there was actually a TMI in place or when a TMI had just terminated. This resulted in a more balanced distribution of the GS class(GS positive class)versus “No GS”(GS negative class), which might help the training process. While JFK and LGA follow very similar distributions, with 40% and 42% GS positive class respectively, EWR has proportionally fewer GS incidents at 28%. Our subsequent phase involves evaluating the performance of both hierarchical structure and independent structure using different state-of-the-art multi-class classifier models such as Random Forest, Decision Trees, K-nearest Neighbors, and Logistic Regression and forecast the duration and scope of the GSs.

Farzan Masrour Shalmani↗