Bridging Experimental Simulations with Computational Frameworks for Time-Resolved Characterization of Fluid-Structure Interactions

Bridging Experimental Simulations with Computational Frameworks for Time-Resolved Characterization of Fluid-Structure Interactions

Çakir, Bora (TU Delft Aerospace Engineering)

van Oudheusden, B.W. (mentor)
Sciacchitano, A. (graduation committee)

Delft University of Technology

Aerospace Engineering

HOMER

2020-09-30

Closure of Collar's triangle represents a complete framework of fluid-structure interactions (FSI) enabling the comprehensive understanding of different design elements compromising aeronautical applications. Experimental methods such as tomographic particle image velocimetry (Tomo-PIV) are proven to provide accurate acquisition opportunities of fluid properties in three dimensional domains. However, not only the closure problem requires simultaneous investigations of fluid and structure behaviors but also the maximum attainable measurement volumes for characterizing these behaviors are severely limited for conventional Tomo-PIV applications. Therefore, a large scale Tomo-PIV setup capable of measuring simultaneous flow motion via Helium Filled Soap Bubbles (HFSB) and structure motion by means of surface markers is employed for experimental investigations of turbulent boundary layer interactions with an unsteadily deforming elastic membrane. Despite the aforementioned benefits of large scale tomographic PIV/PTV techniques, available spatial resolution characteristics for time-resolved flow field measurements are significantly restricted due to the tracer particle specifications of HFSB. This restriction requires additional post-processing algorithms to be applied over the raw experimentally acquired datasets in order to mitigate the effect of experimental trade-off between the temporal and spatial resolution specifications, and allow instantaneous flow field characterization. Although state-of-the-art data assimilation approaches provide the capability of reconstructing high resolution flow features by interpolating the available scattered particle tracking information for global time-resolved flow field reconstruction, drawbacks due to the inability of resolving the viscous effects for near wall flow behavior, incapability of handling physical intrusions to the flow field and propagation of measurement errors, compromise accurate characterization of flow structures in close proximity of the FSI interface. First of all, these algorithms are developed for reconstructing flow features involved with engineering applications where the flow conditions are dominated by turbulent characteristics. However, the fluid behavior in close proximity of walls are dictated by non-negligible viscous forces. Hence, achieving dense interpolation of flow properties of near wall features requires proper characterization of viscous effects for which the thesis proposes the employment of wall function applications for appropriate boundary condition determination. Secondly, even though the available data assimilation methods are able to handle flow behavior around objects, characterization of regions in direct interaction with the object boundaries is not possible. Therefore, in order to handle non-uniform solid boundaries, computational FSI frameworks of the Arbitrary Lagrangian-Eulerian (ALE) and the immersed boundary treatment methods are utilized. Furthermore, as the local closure of Collar’s triangle demands fluid loading over the structural surfaces to be determined, surface pressure information is required to characterize the aeroelastic interactions. Hence, an alternative approach of non-intrusive surface pressure reconstruction from PIV data over unsteadily deforming non-uniform boundaries is introduced via interpreting the ALE method with boundary fitted coordinate systems. Finally, measurement and processing errors contained within the experimental procedures propagate through the data assimilation algorithms. Therefore, to enable the the mitigation of experimental measurement and processing errors, an alternative approach of governing equation based dense flow field interpolation is developed using solenoidal and irrotational radial basis functions (RBF). Capabilities of the proposed methods within the thesis project are demonstrated with various theoretical, numerical and experimental test cases. The wall function application enabled accurate characterization of average streamwise velocity profiles as well as providing slight improvements on the fluctuating velocity components compared to the no-slip boundary condition implementation. Both approaches introduced for handling of physical intrusions resulted in increased coherence levels of flow behaviors within the respective test cases, where the local variations of velocity components revealed promising improvements against the state-of-the-art assimilation algorithms favoring the developed methods in terms of providing greater agreements to the reference flow fields. The introduced surface pressure reconstruction scheme with boundary fitted coordinates yielded relative error levels confined below 4% compared to analytical flow field properties where the resultant errors are computed to be related to the numerical truncation errors rather than the discretizations of mesh deformations or vectorial transformations. The developed alternative approach of dense flow field interpolation with solenoidal and irrotational RBFs allowed inherent mass conservation for the velocity field reconstructions that significantly increased the agreement of the interpolated velocity and vorticity fields with the reference flow field information while irrotationality imposition on the material accelerations provided elevated accuracy levels for pressure field computations. Consequently, the proposed methods of wall function implementation, ALE-VIC+ and ImVIC+ are utilized for experimental characterization of surface pressure variations over the unsteadily deforming elastic membrane exposed to turbulent boundary layer conditions. The instantaneously available low density particle tracking information is assimilated towards dense interpolation of material accelerations to capture temporal evolution of static pressure values at the central membrane location. The superior accuracy specifications of both ALE-VIC+ and ImVIC+ against linear interpolation owing to the kinematic characterization of the influence of membrane motion on the flow field properties provided greater agreements of the non-intrusive time-resolved pressure field computations with the pressure tab measurements.

Fluid-structure interaction
Data assimilation
Pressure from PIV

http://resolver.tudelft.nl/uuid:74e45067-5fe8-4164-a31e-3182529c9583

Student theses

master thesis

© 2020 Bora Çakir