Print Email Facebook Twitter Re-entry vehicle aerodynamic database reconstruction from the analysis of test dynamics Title Re-entry vehicle aerodynamic database reconstruction from the analysis of test dynamics Author Bunt, Riccardo (TU Delft Aerospace Engineering) Contributor Naeije, M.C. (mentor) Sudars, Martins (graduation committee) Degree granting institution Delft University of Technology Programme Aerospace Engineering Date 2019-12-13 Abstract During the design and analysis of a mission characterised by re-entry flight, consistent effort and resources are invested in the definition of the aerodynamic database of the re-entry vehicle. An aerodynamic database collects a set of dimensionless coefficients that describe the interaction of the vehicle's geometry and airflow, and an accurate estimate of its elements is key to the correct modelling of the accelerations experienced during flight.The methodology elaborated herein focuses on the reconstruction of the aerodynamic database of a capsule-shaped vehicle, based on the analysis of simulated data from wind tunnel tests and drop tests. The estimation process is characterised by linear regression of the data that requires linearisation of the dynamics, with the use of Taylor series expansions, and a polynomial representation of the coefficients' dependency on the angle of attack. The preliminary estimate of the coefficients computed by linear regressions is introduced in data smoothing models in order to reduce the noise and errors present in the dataset. The effectiveness of the Extended Kalman filter, the Unscented Kalman filter and the Square Root Unscented Information filter applied to the measurement is established and the results proved their performance to be comparable one to another in the presented problem.Finally, the core of the research performed is related to nonlinear regression methods, oriented towards aerodynamic database reconstruction. At first, an analytical approach is elaborated by defining a harmonic solution for curve fitting of oscillatory dynamics. However, the iterative Gauss-Newton algorithm does not converge to a definition of the regression parameters in the presented case. The focus is therefore then concentrated on optimisation methods. Multi-island Genetic Algorithm, Adaptive Simulated Annealing, Nealder & Mead Downhill Simple, Hooke-Jeves Direct Search and a Hybrid Algorithm are all methods applied to the problem. From the analysis of the results, based on a simulated wind tunnel test of a Hayabusa type capsule in subsonic flow regime, three algorithms emerge for greater accuracy in the reconstructed database: the Hybrid Algorithm, the Hooke-Jeves Direct Search and the Adaptive Simulated Annealing method. Subject AEDB reconstuctionAerodynamic databaseRegression analysisWind tunnel simulationRe-entry vehicleCapsuleDrop test simulationOptimisation Algorithms To reference this document use: http://resolver.tudelft.nl/uuid:0220e4e5-978f-423c-b443-508011f0b13c Embargo date 2021-12-13 Part of collection Student theses Document type master thesis Rights © 2019 Riccardo Bunt Files PDF MSc_Thesis_RiccardoBunt_4757025.pdf 13.85 MB Close viewer /islandora/object/uuid:0220e4e5-978f-423c-b443-508011f0b13c/datastream/OBJ/view