This paper presents a numerical simulation and experiment on the effect of the variable pitch angle on the performance of a small vertical-axis wind turbine (VAWT) with straight blades. The power coefficient of the VAWT was measured in an open-circuit wind tunnel. By conducting two-dimensional unsteady computational fluid dynamics simulations using the RNG k-ε, Realizable k-ε, and SST k-ω models, the power and torque of the VAWT and the flow around the straight blades were also analyzed. The numerical simulation of the power performance results were validated using wind tunnel experimental data. The results of both the numerical simulations and experiments showed that a VAWT with variable-pitch blades has better performance than a VAWT with fixed-pitch blades. The numerical simulation of the performance using the RNG k-ε turbulence model had good qualitative agreement with the experimental results. The numerical simulation was able to capture the flow separation on a blade, and it was shown that a variable-pitch blade can suppress the flow separation on its blades at a tip speed ratio lower than that of fixed-pitch blades.
We present a numerical method based on hierarchical Cartesian grids for the simulation of multiphysics problems, here in particular for the conjugate heat transfer between a fluid and a solid. The data structures developed for this simulation method allow to account for moving objects and are especially well suited for massive parallelization. The major problem of a suitable domain decomposition for a coupled fluid and heat conducting domain is solved by discretizing both domains on a joint hierarchical Cartesian mesh, where the individual cells can be marked according to the underlying physics, i.e. to be a fluid cell, a heat conducting cell or both. Individual solvers for the Navier-Stokes equations and the heat conduction equation are implemented which only operate on the Cartesian cells belonging to the fluid or solid subset of the joint hierarchical mesh. The solution strategy is validated against the analytical solution of the convective heat transfer between a heat-conducting solid flat plate and a laminar incompressible boundary layer. The applicability of the method for moving objects is then demonstrated by solving a conjugate heat transfer problem for a heated and moving cylinder in a laminar flow.
A butterfly wind turbine (BWT) is a kind of vertical axis wind turbine (VAWT) with closed-loop blades. These blades form a double-blade structure, which is expected to improve self-starting properties and reduce energy costs because of their simple construction. Two models of micro BWTs (diameter: 0.4 m; height: 0.3 m) were built and subjected to wind tunnel testing. One of the models had a symmetrical blade section and the other had a cambered blade section with a mean line that followed a curved path in a flow curvilinear relative to the blade. Experimental results showed that the cambered blade rotor was superior to the symmetrical blade rotor in terms of torque and power coefficients at higher tip speed ratios (TSR). However, at low TSRs, the performance of the symmetrical blade rotor tended to be higher than that of the cambered blade rotor. To investigate the effects of blade section on the performance and flow field of the double-blade rotor, two-dimensional computational fluid dynamics (2D-CFD) analysis was carried out for two double-blade rotors with symmetrical and cambered blades. Although 2D-CFD analysis is not suitable for the quantitative performance analysis of the three-dimensional BWT, the CFD results showed the same tendency of the torque and power performance as the experimental results. If the outer blades alone are considered, the cambered blades generate larger torque (or power) than the symmetrical blades at all TSR values, in the case of a large chord-to-radius ratio as with the present rotors. On the other hand, the inner symmetrical blades generate more torque (or power) than the inner cambered blades at TSRs less than 1.5. A TSR of 0.75, at which the symmetrical blade rotor showed the highest torque coefficient, was intensively analyzed in terms of the aerodynamic forces and torques calculated by the 2D-CFD. Under this condition, the inner blade of the symmetrical blade rotor generated positive torque at a wider range of azimuth angles than the cambered blade rotor.
Direct numerical simulation of a turbulent mixing layer with a transversely oscillated inflow is performed. The inlet flow is generated by two driver parts of turbulent boundary layers. The Reynolds number based on the freestream velocity on the low speed side, UL, the 99% boundary layer thickness of the inflow, δ, and the kinematic viscosity, v, is set to be Re = 3000. In order to compare the results with the experimental study of Naka et al. [Naka, Tsuboi, Kametani, Fukagata, and Obi, J. Fluid Sci. Technol., Vol. 5, pp. 156-168 (2010)], the angular frequency of the oscillation was set to be Ωc = 0:83 and 3.85 (referred to as Case A and Case B, respectively). From the three-dimensional visualization, large-scale spanwise vortical structures are clearly observed in the controlled cases. The momentum thickness and the vorticity thickness indicate that the mixing is enhanced in Case A, while it is temporarily suspended in Case B. In both cases, the Reynolds normal stresses are increased in the region right downstream of the forcing point due to the periodic forcing. Furthermore, in Case B, the Reynolds shear stress (RSS), -u´v´, is suppressed in the region downstream of the forcing point. The spatial development of the turbulent energy thickness, δk, and the Reynolds shear stress thickness, δrss, show that the Reynolds shear stress in Case B is decreased by the control despite the increase of the turbulent kinetic energy. From the spectral analysis, large-scale spanwise structures are found to be caused by the periodic forcing, while the spectra of the spanwise velocity fluctuations are nearly unchanged. Co-spectra of the Reynolds stresses show that the present forcings generally enhance the long wavelength component. In Case B, however, the long wavelength component of the Reynolds shear stress is not increased in the downstream region.