Numerical modelling of ultra-thin structures in a fluid domain

Abstract

Folding structures with a large surface area into a compact configuration and deploying them only when required can be achieved by employing origami based folding patterns. This technique is widely seen in nature and mimicking such mechanisms for folding and deployment purposes can lead to highly efficient systems. This study focuses on the biomimicry of insect wings in the design of wings for insect-sized micro air vehicles (MAVs). In order to achieve the required aerodynamic lift and forward propulsion, the wings have to be sufficiently large. At the same time, for manoeuvrability while on ground and also to protect the wings while not in use, the size can be a limiting factor. The wing-folding technique used by certain class of insects is an attractive design solution, which however leads to the need for rigorous numerical analysis, as the fold lines introduced for folding purposes can alter the aerodynamic behaviour of the structure. For a comprehensive analysis a fluid structure interaction (FSI) framework will be required. This thesis presents a detailed study of how the capabilities of commercially available computational structural dynamics (CSD) and computational fluid dynamics (CFD) codes can be exploited in conducting a FSI study of a system involving thin, creased, membranous structures immersed in flowing fluid. For this purpose, initially the suitability of two widely used software environments, ABAQUS and ANSYS Workbench, was investigated. Based on requirements and resource availability ABAQUS CAE and CFD solvers were chosen for the current work and a rigorous study of the theoretical concepts employed by these codes was conducted to ensure that it was in line with the objective at hand. The FSI framework was implemented in three stages. First, the structural domain was developed. A Mooney-Rivlin material model with a membrane formulation was found to be suitable for modelling thin biological membranes. However, for simplicity a linear elastic material with plastic properties was initially assumed and was used in combination with a shell formulation. iii Next, the fluid domain was developed using a Spalart-Allmaras turbulence model and was verified qualitatively based on the variation in flow around a rigid cylinder with Reynolds Number. Finally, both aforementioned domains were coupled using an explicit Gauss-Seidal scheme through the Abaqus Co-simulation Engine. This was verified against a standard benchmark study conducted by Turek and Hron (2006), which involves a cylinder immersed in a laminar flow with a thin, flexible flap attached to its downstream side. The sinusoidal oscillation of the flexible flap was captured and features such as vortices and streamlines were reproduced to an acceptable extent. However, the numerical accuracy of the solutions needs further improvement. As the next stage, the behaviour of a creased membrane immersed in the above fluid domain was studied. To introduce the crease effect, the crease geometry was pre-defined on the structural component and this was then coupled with a fluid domain set up following the same procedure as in the previous case. This method, while presenting a simpler modelling path, has the disadvantage that the exact hinge-effect introduced by an actual crease-line is not captured. However, the alteration to the flow path from a geometric perspective was captured. In the developed model a main concern was the processing time. To address this issue a two-dimensional shell formulation was used in this work. The presence of the shell was indicated by a seam crack in the fluid domain. Despite that, the model still suffered from the severe disadvantage of processing time. Thus in future works, other techniques should be incorporated to overcome this constraint. Also, the actual behaviour of the crease-line and the region close to it should be captured. A technique that could be used in future works was presented in the final section of the thesis.