Person:

Mavris, Dimitri N.

Permanent Link

https://hdl.handle.net/1853/71501

Associated Organization(s)

Organizational Unit

Daniel Guggenheim School of Aerospace Engineering

Organizational Unit

Aerospace Systems Design Laboratory (ASDL)

Full item page

Publication Search Results

Now showing 1 - 10 of 49

Machine Learning Enabled Turbulence Prediction Using Flight Data for Safety Analysis

(International Council of the Aeronautical Sciences (ICAS), 2021-09) Emara, Mariam ; dos Santos, Marcos ; Chartier, Noah ; Ackley, Jamey ; Puranik, Tejas G. ; Payan, Alexia P. ; Kirby, Michelle R. ; Pinon, Olivia J. ; Mavris, Dimitri N.

The hazards posed by turbulence remain an important issue in commercial aviation safety analysis. Turbulence is among the leading cause of in-flight injury to passengers and flight attendants. Current methods of turbulence detection may suffer from sparse or inaccurate forecast data sets, low spatial and temporal resolution , and lack of in-situ reports. The increased availability of flight data records offers an opportunity to improve the state-of-the-art in turbulence detection. The Eddy Dissipation Rate (EDR) is consistently recognized as a reliable measure of turbulence and is widely used in the aviation industry. In this paper, both classification and regression supervised machine learning models are used in conjunction with flight operations quality assurance (FOQA) data collected from 6,000 routine flights to estimate the EDR (and thereby turbulence severity) in future time horizons. Data from routine airline operations that encountered different levels of turbulence is collected and analyzed for this purpose. Results indicate that the models are able to perform reasonably well in predicting the EDR and turbulence severity around 10 seconds prior to encountering a turbulence event. Continuous deployment of the model enables obtaining a near-continuous prediction of possible future turbulence events and builds the capability towards an early warning system for pilots and flight attendants.
Impact of Adverse Weather on Commercial Helicopter Pilot Decision-Making and Standard Operating Procedures

(Georgia Institute of Technology, 2021-08) Speirs, Andrew H. ; Ramee, Coline ; Payan, Alexia P. ; Mavris, Dimitri N. ; Feigh, Karen M.

Helicopter pilots face unique challenges with regard to adverse weather when compared to fixed-wing pilots. Rotorcraft typically operate at lower altitudes in off-field areas that are not always well covered by weather reporting stations. Although recent technological advances have increased the amount of weather data that pilots can access in the cockpit, weather remains a factor in 28% of fatal helicopter accidents. In this work, commercial helicopter pilots were surveyed and interviewed to better understand how they gather and process weather information, what the perceived limitations of current weather tools are, and how their decision-making process is affected by the information they gather and/or receive. Pilots were found to use a wide variety of weather sources for their initial go or no-go decision during the preflight phase, but use fewer weather sources in the cockpit while in-flight. Pilots highlighted the sparsity and sometimes inaccuracy of the weather information available to them in their prototypical operational domain. To compensate, they are forced to rely on local and experiential weather knowledge to supplement weather reports while still working to mitigate other external pressures. Based on the literature and on results from this work, recommendations are made to address the weather-related gaps faced by the rotorcraft community. This includes the installation of additional weather reporting stations outside of airports and densely populated areas, the further promotion of the HEMS tool to helicopter pilots in all industries, the development of weather tools capable of visualizing light precipitation such as fog, and the development of in-flight graphical displays that can help reduce the cognitive workload of interpreting weather information.
Performance Assessment of a Distributed Electric Propulsion System for a Medium Altitude Long Endurance Unmanned Aerial Vehicle

(Georgia Institute of Technology, 2021-08) Markov, Alexander A. ; Cinar, Gokcin ; Gladin, Jonathan C. ; Garcia, Elena ; Denney, Russell K. ; Mavris, Dimitri N. ; Patnaik, Sounya S.

Distributed propulsion systems are enabled by electrified aircraft and can provide aero-propulsive benefits. The magnitude and impact of these benefits rely on the location of propulsors on the vehicle, the amount of propulsive force generated by those propulsors, vehicle geometry, and the extent of hybridization of the propulsion system. With an increased number of degrees of freedom over conventionally electrified aircraft, the full extent of the impacts of this technology have not yet been explored, especially for military applications. This study builds on a previous one that implemented a series hybrid and turboeletric propulsion architecture on a turboprop UAV, by introducing a distributed electric propulsion system on the same vehicle. The previous study showed that with a hybrid architecture, the same performance, in terms of range and endurance, could not be achieved for a fixed gross take-off weight. This study investigates the impact of the distributed propulsion system with the goal of identifying the benefits over the previous vehicle and determining the level of technology required to break even with the conventional propulsion UAV. In incorporating the new propulsion system, the engine and main motor are resized, leading edge wing mounted propellers and motors are added to the configuration, and a new battery sizing strategy is implemented. Preliminary results show that, although this new system shows increased range and endurance over the series hybrid vehicle, it still falls short compared to the conventional vehicle with current levels of technology. Although improvements are needed to the electrical system technology to reduce the weight enough to break even with the conventional system, the new vehicle shows increased performance during climb, and has the capability to store energy during the mission. With the proper power management and battery utilization strategies, this system can provide reduction in fuel burn and improved performance during certain phases of the mission which could be beneficial for military applications.
Analysis of Weather-Related Helicopter Accidents and Incidents in the United States

(Georgia Institute of Technology, 2021-08) Ramee, Coline ; Speirs, Andrew H. ; Payan, Alexia P. ; Mavris, Dimitri N.

Helicopters typically operate at lower altitudes than fixed-wing aircraft and can take-off and land away from airports. Thus, helicopter pilots have decreased access to weather information due to connectivity issues or sparsity of weather coverage in those areas and at those altitudes. Moreover, regulations allow most rotorcraft to operate in marginal weather conditions. Therefore, weather is a challenge to rotorcraft operations. In this study, rotorcraft events in the United States between 2008 and 2018 in which weather was determined to be a factor are analyzed using the National Transportation Safety Board aviation database. Results show that weather was a factor in 28% of rotorcraft fatal accidents. Wind was involved in most incidents but more rarely involved in fatalities. Bad visibility conditions due to a combination of low illumination and clouds were responsible for most fatal weather-related accidents. Personal flights had the highest accident and incident rates. Finally, the Helicopter Air Ambulance industry had the largest number of incidents and accidents related to visibility conditions out of all other industries. The authors recommend improving awareness of the conditions in which weather events occur and improving training to maintain control of the aircraft in windy conditions or during inadvertent instrument meteorological conditions.
Modeling and Simulation of Novel Electric/Hybrid Electric Multicopter Architectures for Urban Air Mobility

(Georgia Institute of Technology, 2021-08) Demers Bouchard, Etienne ; Verberne, Johannes ; D’Arpino, Matilde ; Ozcan, Metin ; Porpora, Francesco ; Gladin, Jonathan C. ; Patel, Srujal ; Justin, Cedric Y. ; Mavris, Dimitri N.

This paper introduces a dynamic simulation environment developed for novel multi-copter aircraft architectures. The development is motivated by the need to better understand the safety implications of architectural design choices and to provide a formal reliability assessment framework for new Vertical Take-Off and Landing (VTOL) concepts able to consider various airframe and subsystems dynamic behavior. The concepts of interests are different multi-copters configurations investigated by NASA and featuring either electric, hybrid electric, or turboshaft driven powertrains. The simulation environment is a timemarching dynamic simulator formulated using physics-based subsystem models for the batteries, electric motors, turboshaft engines and electric generators. Identified fault modes are integrated into the subsystem models for subsequent use during reliability assessments. The impacts of subsystem faults are propagated to the vehicle flight dynamic response for analysis of their impact on the ability of the vehicle to sustain safe operations. Detailed features of the electric quadrotor model are provided to illustrate the simulation capabilities. Some faults are inserted on the different aircraft in hover and the subsystems behavior is successfully propagated at the vehicle level.
Imitation Learning for UAS Navigation in Cluttered Environments

(Georgia Institute of Technology, 2021-01-04) Harris, Caleb M. ; Choi, Youngjun ; Mavris, Dimitri N.

Autonomous navigation is a critical component for the use of unmanned aerial systems (UAS) in complex tasks such as package delivery and disaster response. In recent years, these systems have seen increased usage in harsh tasks such as search and rescue and disaster relief, however there remains challenges for efficient and safe operation in a fully autonomous mission. This work seeks to provide a data-driven, vision-based method to navigating and searching through a clustered environment, which is high-speed, low-cost and vehicle-agnostic. This is done by first assuming obstacle avoidance as a two-dimensional navigation task that can be solved by knowing the relative location of the goal and the 2D image of the obstacle in the camera frame. Imitation learning is used to train a deep neural network from an expert planning policy, while a model predictive controller tracks the target. All the processing is capable onboard the vehicle, with the assumption that the general target direction is forward of the camera-frame, and that the global state estimation error is low. The framework and trained model are tested in simulation, with a quadcopter conducting search scenarios in different environments. The resulting framework is quicker to avoid obstacles and can be applied on small, low-cost systems with a single monocular camera.
Use of Machine Learning to Create a Database of Wires for Helicopter Wire Strike Prevention

(Georgia Institute of Technology, 2021-01-04) Harris, Caleb M. ; Achour, Gabriel ; Payan, Alexia P. ; Mavris, Dimitri N.

Rotorcraft collisions with wires and power lines have been a major cause of accidents over the past decades. They are rather difficult to predict and often result in fatalities. For this reason, there is a push to provide pilots with additional information regarding wires in the surrounding environment of the helicopter. However, the precise locations of power lines and other aerial wires are not available in any centralized database. This work proposes the development of a wire database in two phases. First, power line structures are detected from aerial imagery using deep learning techniques. Second, the complete power grid network is predicted using a centralized many-to-many graph search. The two-step framework produces an approximate medium-voltage grid stored as a set of connected line segments in GPS coordinates. Experiments are conducted in Washington D.C. using openly available datasets. Results show that utility pole locations can be predicted from satellite imagery using deep learning methods and a full grid network can be generated to a level of detail depending on computational power and available data for inference in the graph search. Even with limited computational resources and a noisy dataset, over a a fourth of the grid network is directly predicted within a range of seven meters, and the majority of the network is visually inferred from nearby detections. Moving forward, the goal is to apply the proposed framework to larger regions of the U.S., with rural and urban environments, to map all wires and cables that are a threat to rotorcraft safety.
Parametric Design, Manufacturing and Simulation of On-Demand Fixed Wing UAVs

(Georgia Institute of Technology, 2021-01-04) Chitale,Yash H. ; Justin, Cedric Y. ; Mavris, Dimitri N.

As the market for Unmanned Aerial Vehicles (UAVs) continues to expand, an unfulfilled need has been identified for tailor-made solutions leveraging an end-to-end process for the design and manufacture of the vehicle. The use of computer aided design combined with new manufacturing techniques allows small UAVs to be parametrically sized and quickly prototyped and deployed. This parametrization technique can be used throughout the entire design process to create optimized, attritable, on-demand solutions that can be adapted to evolving customer requirements. High-level requirements are mapped to quantitative design constraints and an automated process uses these constraints to design and manufacture a vehicle within a specified amount of time. The proposed framework is demonstrated with the generation of a fixed wing UAV solution for the detection and tracking of wildlife in remote areas. National Parks seek to prevent illegal poaching but often lack either the resources to monitor endangered animals, or the budget to purchase UAVs specially designed for wildlife tracking. First, mission requirements are identified and define a design space from which an optimal design point is selected. This design point sizes a UAV model, which is then optimized to minimize manufacturing time with the objective to yield a ready-to-fly solution within 48 hours. A flight simulation of the mission is then performed to ensure that the vehicle will fly as designed. Structural limitations of the UAV are accounted for and linked to parameters of the flight control algorithm to ensure that the UAV can safely fly its mission.
Evaluation of Off-Nominal Performance and Reliability of a Distributed Electric Propulsion Aircraft during Early Design

(Georgia Institute of Technology, 2021-01-04) Bendarkar, Mayank ; Sarojini, Darshan ; Harrison, Evan D. ; Mavris, Dimitri N.

General Aviation (GA) is likely to be at the forefront of a paradigm change in aviation, where the introduction of novel concepts such as Urban Air Mobility (UAM), architectures like e-VTOL, and technologies like Distributed Electric Propulsion (DEP) are expected to make aircraft more efficient and reduce their environmental footprint. However, these architectures carry with them an uncertainty related to the off-nominal operational risk they pose. The limitations and off-nominal operational considerations generally postulated during traditional safety analysis may not be complete or correct for new technologies. While a lot of the literature surveyed focuses on improving traditional methods of safety analysis, it still does not completely address the limitations caused due to insufficient knowledge and experience with transformative technologies. The research objective of the present work is to integrate the Bayesian safety assessment framework developed previously by the authors with conceptual and 6-DoF performance models for DEP aircraft to evaluate off-nominal performance and reliability using information that is typically available in conceptual or preliminary design phases. A case study on the electric power architecture of the the NASA Maxwell X-57 Mod. IV is provided. A maximum potential flight path angle metric, as well as trimmability considerations using a 6-DoF model constructed using available literature help determine hazard severity of power degradation scenarios. Bayesian failure rate posteriors are constructed for the different components in the traction power system, which are used in a Bayesian decision framework. The results indicate that while most of the components in the traction power architecture of the X-57 Mod. IV are compliant with failure rate requirements generated, the batteries, cruise motors, and cruise motor-inverters do not meet those requirements.
Improving Courier Service Network Efficiency through Consolidations

(Georgia Institute of Technology, 2021-01-04) Zhang, Wenxin ; Payan, Alexia P. ; Mavris, Dimitri N.

Service network design is a significant consideration for courier companies because an efficient design reduces operating costs while maintaining service quality. While companies typically rely on subject-matter experts knowledge to modify their service network design on a regular basis based on changes in demand, some of them have also developed an optimization-driven approach to improve the design of their service network in the long-term. Typically, service networks are based on a hub-and-spoke design. However, operating costs may be reduced by adding consolidations on the pickup and/or the delivery routes into and out of hubs. Consolidations are locations where packages can be aggregated from multiple spokes to go into a hub or can be disaggregated to be delivered to multiple destinations from a hub. This service network design feature ultimately reduces the number of aircraft used on each route and therefore decreases the operating costs. In this study, we use Integer Programming with hierarchical objectives to generate consolidation options. The proposed algorithm accounts for network-wide demand considerations and aims at reducing costs from operating several modes of transportation by minimizing the number of consolidation locations while ensuring that every package is served and gets delivered on time at its intended destination. The algorithm is being implemented on the entire domestic U.S. market and has the flexibility to generate one or more consolidation options for each group of packages going from a given origin to a given destination. Results from the optimization are compared to solutions from a heuristic approach based on a series of geographical and operational rules. Results show that the optimization approach is able to generate better consolidation options compared to the heuristic approach. In particular, allowing packages to consolidate at a maximum of three consolidation locations results in a two percent reduction in the total costs over individual days of operations, and in nearly a one percent reduction in the total costs over a week of operations, for similar computational times. Although these reductions seem small, operating costs for courier companies tend to be in the millions or billions of dollars. Therefore, even a one percent reduction is significant.