The CEREAL lab focuses on a variety of research topics, with computational work, experimental work, and hybrid work that combines both. Below, you can find more information on a selection of specific projects being undertaken and the people who work on them.
Computational
Mid-Fidelity Lattice-Boltzmann Rotorcraft Aerodynamics Simulations
Among other projects, the CEREAL lab works with ONR and the U.S. Navy on the Lattice-Boltzmann Real-time Dynamic Interface (LBRDI) project. The aim of this project as a whole is to develop GPU accelerated Lattice-Boltzmann Method (LBM) based simulations as a mid-fidelity tool for efficient aerodynamic analysis of vertical lift vehicles, the interactions of these vehicles with complex flow environments and/or solid boundaries, and entire mission scenarios. These LBM simulations will accurately predict the coupled ship/aircraft interactions in a real-time simulation environment. Coupled to aircraft flight mechanics, the models are used in piloted flight simulators to improve and de-risk pilot training and to potentially help with simulation-based certification. LBM simulation results are compared with high-fidelity CFD results and experimental data for validation.
The video below shows LBM simulation results for a full-scale notional UH-60 helicopter operating in the wake of the NATO Generic Destroyer simplified ship geometry. The LBM simulation has been coupled to the Navy’s Example Helicopter (ExHel) code, a validated UH-60 flight dynamics and simulation model. The ship is modeled with a Grad immersed boundary scheme, which allows for ship motion to be incorporated into the simulation. The simulations are conducted using “two-way coupling,” in which both the rotor wake and ship airwakes are modeled within the LBM simulation. This allows for interactional effects between the ship airwake and rotor wake to be resolved and characterized. Indeed, recirculation and ground effects are evident when the helicopter hovers over the ship landing deck; these effects could not be effectively resolved in a one-way coupled simulation. Using GPU acceleration, simulations are real-time capable, which will allow the two-way coupled LBM inflow model to be used in piloted flight simulation applications.
Other research focuses on the development of a GPU-accelerated Lattice Boltzmann Method (LBM) solver as a mid-fidelity tool for aerodynamic analysis of multirotor vertical lift uncrewed aerial vehicles (UAV). This project is sponsored by the Army Research Lab (ARL) and uses GPU-accelerated LBM as an efficient aeromechanics analysis tool for UAVs, simulations of entire mission task elements and mission scenarios, including associated energy requirements for (electric) vehicles in steady and unsteady flight conditions. As part of the project, the experimental data measured by ARL (forces/moments and flow fields) on generic quadcopter airframes are used to validate the results from the GPU-accelerated LBM solver, find optimal simulation parameters, and refine the numerical models with the aim of getting a quick turnaround engineering answer/solution for UAV and other vehicles.
Personnel: Shreyas Ashok, Erk Kurban, Pyae (Teddy) Su, Duncan Waanders, Braulio Vera Garcia
Tough and Ecological Supercritical Line Breaker for AC (TESLA)
Sulfur hexafluoride (SF6), is a man-made greenhouse gas with a global warming potential 23,900 times that of CO2 and is a key component in high-voltage circuit breakers. The TESLA project aims to reduce greenhouse gas emissions by using supercritical CO2 instead of SF6 as an arc quenching medium. The work CEREAL is contributing to the project is to investigate the fluid dynamics inside the circuit breaker and provide guidance for mechanical designs of a high-pressure tank, contact system, and arc quenching mechanism.
Personnel: Zoelle Wong
Experimental
Rotor—Rotor Aerodynamic Interactions
CEREAL’s experimental work in the area of rotor–rotor aerodynamic interactions focuses on evaluating how the interactions affect the aerodynamics and performance metrics of the individual rotors and of the overall vehicle. The experiments are performed in Georgia Tech’s John J. Harper Wind Tunnel and focus on various parameters. Some such parameters of interest include the configuration of the rotors (e.g. dual and quad rotors), the hub-to-hub spacing of the rotors, flight in confined spaces (e.g. ground, wall, and ceiling effects), and flight in urban environments. These experiments are compared to computational work also done in the CEREAL lab, as well as to flight tests performed by The Ohio State University.
Personnel: Abraham Atte
Propeller–Wing Aerodynamic Interactions
CEREAL’s experimental work in the area of prop–wing aerodynamic interactions focuses on characterizing the flow field around a propeller–wing setup using particle image velocimetry (PIV), propeller and wing loads, and pressure data. Various conditions are tested in Georgia Tech’s John J. Harper Wind Tunnel with the ultimate goal of encapsulating the entire range of flight regimes a tiltrotor VTOL or urban air mobility (UAM) vehicle might encounter. The work includes multirotor configurations, transition regimes for tiltrotors, and dynamic changes in flight parameters.
-10° Wing Pitch
20° Wing Pitch
Personnel: Shreyas Srivathsan
Near-boundary Aerodynamics on Multirotor Systems
The growth of the Unmanned Aerial Vehicle (UAV) industry is outpacing our understanding of how UAVs behave in near-boundary environments. Search and rescue or surveillance UAV operations often occur in tight, confined spaces filled with complex obstacles and boundaries. Near-boundary flight can provide aerodynamic benefits, such as the ground effect used by animals or rotorcraft. However, near-boundary flight advantages are difficult to harness because boundary effects may also have detrimental and destabilizing impact on the vehicle flight dynamics. These effects perturb thrust and lift (near wall–air or water–air boundaries). Therefore, CEREAL studies the aerodynamics and performance of multirotor systems operating in the complex flow environment near solid structures and ceiling/ground/corners in order to explore these near-boundary effects.
Personnel: Abraham Atte
Ship Airwake–Rotor Interactions
The shipboard operation of rotorcraft poses significant challenges to the pilots onboard or the flight dynamics and controls of unmanned aerial systems (UAS). Aerodynamic interactions between the ship’s airwake and the wake of the rotor(s) result in a highly unsteady turbulent flow field that impacts the rotorcraft aerodynamic and flight dynamic behavior. Although researchers have been studying this topic for several decades, there are still challenges and unresolved questions to be tackled. CEREAL is investigating this further through model-scale experiments conducted in Georgia Tech’s John J. Harper Wind Tunnel along with partners at Embry-Riddle Aeronautical University (ERAU). This project aims to provide a holistic understanding of the standalone ship airwake, and the airwake–rotor interaction through data collected using particle image velocimetry (PIV), surface pressure measurements, and 6-degrees-of-freedom rotor force measurements. Flow physical insights and measured data are used to verify and validate the numerical simulations developed under computational projects within the lab.
Personnel: Wei-Han Chen
Electric Ducted Fan (EDF) Characterization and Aero-Propulsive Interactions for UAM/RAM Applications
Distributed electric ducted fan propulsion systems have captured the interest of many urban air mobility/regional air mobility (UAM/RAM) vehicle manufacturers as they provide a viable way to make their aircraft completely electric while also increasing propulsive efficiency and reducing noise. Some examples are the ONERA Dragonfly and the Lillium Jet. CEREAL’s research focuses on characterizing isolated EDFs as well as different wing integrations using wind tunnel tests. The integration studies focus on how the EDF performance changes with each integration using its isolated version as a baseline comparison. In collaboration with project partners, comparisons of the isolated thrusters are made with flight tests using an EDF-equipped flight test vehicle. Using the experimental database, computationally inexpensive aerodynamic design tools are validated for this use case, and then used to predict propulsion integration and aero-propulsive interaction effects that may not be studied experimentally.
Personnel: Dr. Sihong Yan, Bryan Regan
Adaptive Flutter Suppression
Typical active flutter suppression controllers rely on inertial and elastic data feedback such as acceleration and strain. This approach has been shown to be effective, however potential improvements may be made with the inclusion of additional information in the form of surface pressure data. Wind tunnel testing is conducted on a cantilevered wing model instrumented with an angular rate sensor, an accelerometer, and surface pressure transducers. Trailing edge flaps on the wing were used as the control effectors. The open-loop behavior was characterized, then control with angular rate feedback into an adaptive controller was evaluated. Multiple configurations of inertial and surface pressure feedback control were evaluated, with the final configuration achieving a 25% increase in freestream velocity over the open-loop case and 10% over the angular rate feedback closed-loop case. This was found to be a valid approach and did not rely on precise modelling of the entire test rig. Overall, it was shown that the inclusion of surface pressure data provided information not present in the inertial data which enhanced the closed-loop stability of the test rig.
Personnel: Jacob Szymanski, Kenjiro Lay
Hybrid
Rotor—Rotor Aerodynamic Interactions
CEREAL’s work in the area of rotor–rotor aerodynamic interactions focuses on evaluating how the interactions affect the aerodynamics and performance metrics of the individual rotors and of the overall vehicle. An advanced adaptive Cartesian grid-based CFD solver suited for computationally efficient flow prediction over complicated real-life configurations is used to analyze dual- and quad-rotor configurations with varying hub-to-hub spacing and forward flight speeds. These computational results are compared to experimental wind tunnel tests (also done in the CEREAL lab) and flight tests (done at The Ohio State University).
Personnel: Daley Wylie
AI/ML Tools for Aeromechanics Predictions of Next-Generation Vertical Lift Aircraft
CEREAL’s goal for this project is to use AI/ML tools to develop aeromechanics models for vertical lift aircraft, including next-generation configurations featuring multiple rotors. The models are being developed for: (1) estimation of rotor blade loads from measured blade deformation, and (2) real-time flight dynamics simulations using an efficient mapping between flow field and rotor loads. The research is being carried out in close collaboration between research groups here at Georgia Tech (PI: Prof. Juergen Rauleder) and at the University of Texas at Austin (PI: Prof. Jayant Sirohi).
Personnel: Howon Lee
Large Airfoil Model
Access LAM here
Large language models (LLMs) are rapidly transforming the way people are learning and understanding complex topics. Their capability to be fine-tuned to answer questions on specialized topics suggests the viability of a “large airfoil model (LAM),” which can answer complex technical questions on airfoil aerodynamics and be helpful in the (aircraft, rotorcraft, or wind turbine) wing design process. Such a model requires a rich historical dataset it can easily access. For gaps within the database, the model must be able to generate physic-based responses. CEREAL is working on the development of the main components of the LAM: (1) an airfoil surface pressure information repository of experiments (ASPIRE), a database of digitized pressures measurements sourced from publicly available experimental reports, and (2) a deep airfoil prediction tool (ADAPT), capable of predicting airfoil surface pressure distributions including their uncertainties, and integrated force and moment coefficients. By taking a Bayesian approach in regression, the model can take into account experimental uncertainties during the training process and output the confidence intervals of the model predictions. The model employs a Deep Kernel Learning architecture, combining deep learning for active space mapping and a probabilistic method for uncertainty characterization. The LAM is a component of an interactive aerodynamics tool currently under development which allows users to perform complex aerodynamic design tasks (forward and inverse problems, 3D air load predictions, etc.) without the need for in-depth technical knowledge.
Personnel: Howon Lee