Aeroelasticity of Membrane Kites: Airborne Wind Energy Applications

The climate actions defined by United Nations require a rapid transition to low environmental footprint technologies. The energy sector is the major emitter of carbon dioxide emissions and a significant contributor to extracting resources for fuel and power plant construction materials.
Wind energy is projected to produce a significant share of electricity and energy in the following decades. The wind turbines have a small footprint during the operation, but the turbine with its foundation is a massive structure with a significant material footprint. Airborne wind energy uses tethered devices to harness high-altitude wind energy, substantially reducing bulk material use. However, better models are required to make the systems reliable and efficient.

This thesis focuses on membrane traction kites that harness wind energy by flying fast crosswind maneuvers. A high-fidelity aeroelastic model for the kites is developed to predict the aerodynamic loads and the structural deformations of real systems. The aeroelastic model assumes that the membrane kite flight can be modeled as multiple steady-states without memory from the past. The steady-state aerodynamics are simulated by solving the incompressible Reynolds-averaged Navier-Stokes equations numerically. High-quality numerical grid generation strategies are developed for the unconventional wing shape of the membrane kites.

The membrane kites are tensile structures, and therefore a finite element model with cable and membrane elements without rotational degrees of freedom is used to calculate the deformed shape. The solver calculates the average surface without wrinkles and applies an additional model when an element is under compression. The steady-state response of the structure is calculated with a dynamic relaxation technique. The two solvers are coupled in a partitioned manner, and during each iteration, both solvers compute a steady state. The staggered approach requires several coupling iterations to converge. The fluid mesh needs to be altered to the deformed geometry during each iteration, and therefore, the mesh is deformed with radial basis function with greedy point selection.

This thesis presents three computational studies with the framework. The first two studies focus on the aerodynamics of rigidized LEI kite airfoil and wing. The aerodynamic model is validated with an already existing wind tunnel experiment on a similar airfoil. Generally, the largest model uncertainty in CFD is the mesh and therefore, the uncertainty is assessed by mesh refinement studies. A range of flight conditions is simulated by varying the inflow angle of attack, sideslip angle and Reynolds number. The flow around the wing is characterized by a recirculation zone behind the leading edge tube due to the lack of second skin. The zone is highly influenced by the inflow conditions. The effect of the chordwise inflatable tubes on aerodynamics is assessed by creating a model with and without them. The results show that the chordwise tubes have an almost negligible impact on the aerodynamic forces, which suggests they could be left out of the aerodynamic model in future work, simplifying the mesh generation and mesh deformation.

The third study shows the aeroelasticity of a ram-air kite for several power configurations by changing the trim of the bridle lines. The kite forms a typical ram-air shape with ballooning in between ribs, and the nose of the wing is flattened at the stagnation region. The aerodynamics of the flexible kite is compared to a rigidized version of it. The wing is fixed at the symmetry plane and fixed to the pre-inflated shape with stagnation pressure. The results show that the flexible kite is aerodynamically more efficient than the rigidized version. The morphing wing adapts itself to the incoming flow in a way that extends the range of feasible flight conditions and improves efficiency. The aeroelastic framework converges satisfactorily with all the power setups, and it is computationally relatively inexpensive for fidelity. Consequently, the framework could be integrated into a membrane kite design process and could be a valuable asset in evaluating kite designs.