[Back]

S. Schoder:
"Aeroacoustic Analogies based on Compressible flow data";
Supervisor, Reviewer: M. Kaltenbacher, C. Schram, C. Munz; Maschinenwesen und Betriebswissenschaften, 2019; oral examination: 2019-01-23.

The thesis in hand focuses on a novel numerical simulation method to compute aeroacoustic analogies
based on compressible flow data by a hybrid technique. Industrial applicability of aeroacoustic simulation
technologies is computational demanding. The computational workload is reduced with the hybrid approach
to an efficient minimum. With the proposed workflow we are capable of combining the properties
of a fully resolved compressible flow simulation (including feedback from acoustics to flow structures) and
the desirable advantage of a separated acoustic simulation. A separation of the physical fields during the
simulation yields in a computationally efficient algorithm, which is capable of including relevant physical
effects due to the flow and acoustically specific boundaries, like impedance, can be applied.
In this sense, we extend the hybrid approach from underlying incompressible flow simulations to
compressible flow simulations using Helmholtz projection to obtain a vortical base flow and apply the
known hybrid methodologies. The application of this hybrid methodology seems to be unconventional
and fluid dynamically not rigorous, but with the correct wave operator the equation obeys the fluid
dynamic conservation equations. We apply the method to aeroacoustic examples involving aeroacoustic
feedback mechanisms, which require a compressible flow simulation. However, practical applications show
that sometimes even for incompressible flow simulations "typical feedback mechanism", as described by
Rossiter, are captured. A short mathematical explanation, why feedback is even possible for incompressible
flow structures, is given based on compact acoustics.
Hybrid aeroacoustic analogies rely on energy conserving and accurate transformation schemes that
convert the known physical quantities, like pressure, and velocity, form one grid to another. Simple
Nearest Neighbor mappings are not accurate enough for source term computation. Therefore, a combination
of a local Radial Basis Function framework and conservative integration procedure relying on cell
intersections is applied to transform the physical quantities and construct accurate derivatives on them
for a robust simulation workflow.