SCCS Colloquium - Oct 17, 2019
Date: | October 17, 2019 |
Room: | 00.08.053 |
Time: | 15:00 - 16:00 |
Manuel Dell'Antonio: Definition and Evaluation of a Flight Control System for Future VTOL Applications
This is a Master's thesis submission talk. Manuel is advised by Julian Rhein (TUM Inst. Flight System Dynamics) and Prof. Hans-Joachim Bungartz.
Today’s vision of urban air mobility includes also Vertical Take-Off and Landing [VTOL] vehicles. New aircraft architectures are established with plans to achieve Entry Into Service [EIS] within the next years. Since new aircraft are in preparation also certification authorities (e.g. EASA, FAA) are currently working on additional regulations for VTOL (e.g. SC-VTOL-01 from EASA). With the planned use cases new flight control system architectures are needed for control of altitude, attitude and propulsion. The objective of this Master Thesis is to develop different flight control system architectures for VTOLs, validate these architectures regarding existing expectations and current regulation objectives and provide a guideline for system architecture definition. Possible protection and detection methods shall be assessed within the study of architectures to get realistic means to comply with the safety requirements.
Keywords: Flight Control System, VTOL, Safety
Nicol Fottner: Developing and Benchmarking a Molecular Dynamics Simulation using AutoPas
This is a Bachelor's thesis submission talk. Nicol is advised by Fabio Gratl.
The C++ library AutoPas is meant to be included into simulation programs, and delivers optimal node-level performance for N-body problems with dynamic auto-tuning at run-time. In this thesis, a molecular dynamic simulation program was built utilizing the flexibility of AutoPas. It provides multiple methods and functionalities in order to create different molecular dynamic scenarios and was evaluated by simulating the spinodal decomposition phenomenon on the CoolMUC-2 cluster. The results show, that the performance of pairwise force calculations increases by exclusively using Structure of Arrays and AVX intrinsics. In general, the performance is inferior to ls1-mardyn.
Keywords: AutoPas, Spinodal decomposition, HPC, Benchmarking