Detached Eddy Simulation for Aerospace Applications

Doctoral thesis, 2021

With the continuous growth in air traffic that we see nowadays, comes an increase in the requirements needed to be satisfied in order to certify an aircraft for operation. These stricter regulations affect aspects such as CO2 emissions, sound pollution and so on, pushing manufacturers to aim for lighter, more efficient, more robust designs. These improvements might be achieved in two different ways; by improving/optimizing existing technology, or by developing new technological concepts. In either of the two scenarios, numerical tools, such as optimization methods or reliable fluid flow simulations play a paramount role.

In this thesis, new functionalities implemented into the in-house compressible Computational Fluid Dynamics (CFD) solver, G3D::Flow, are described. These new additions have been put in place with the objective of performing turbomachinery simulations using hybrid RANS/LES methods as well as nozzle flow simulations. Some of the additions to G3D::Flow include: phase-lagged pitch-wise and rotor-stator interfaces based on the chorochronic method as well as a method based on Proper Orthogonal Decomposition (POD), sliding grid interface and synthetic turbulence injection. The added capabilities, enable G3D::Flow to perform high-fidelity turbomachinery CFD simulations, which were not affordable before due to their high computational cost, since truncated domains can be used.

A hybrid RANS/LES simulation of the VOLVO S6 nozzle contour operating under overexpanded conditions is performed. This same geometry, under the same conditions, was previously simulated and reported using a different hybrid RANS/LES methodology. A reduction of over $50\%$ in the difference between the predicted standard deviation of the side loads and those measured in a previous experimental is observed in the current simulation.

In this work, an optimization framework called HAMON is also presented, which is based on evolutionary algorithms. In cases where the optimization is based on computationally heavy tasks, such as 3D CFD simulations, meta-modeling techniques can be used to speed up the optimization processes. HAMON can be used to fine tune an existing design, or as it has been used here, as black-box approach. It has been able to design counter rotating open rotors with more than acceptable performance where no knowledge about propeller aerodynamics was assumed, giving all the design variables more freedom than probably needed. This black-box approach might be specially useful when optimizing new technologies for which no prior knowledge exist, allowing not only to, hopefully, find good designs but also to show the trends of what a good design should be like.