The direct torque control (DTC) and direct force control (DFC) method were introduced to reduce the torque and levitation force ripple in single-winding bearingless switched reluctance motors (SWBSRMs). However, it still has some disadvantages. Firstly, the flux-linkage control is not suitable for the DTC method in SWBSRMs. On the one hand, it can increase the torque ripple. On the other hand, the RMS current can be increased and then the torque-ampere ratio is decreased. Secondly, the vectors selection is also unreasonable, which can increase the torque ripple further. In order to solve these problems, an improved control method based on DTC and DFC method for SWBSRMs is proposed in this paper, which can obtain high torque-ampere ratio and low torque ripple simultaneously. In the proposed method, the flux-linkage loop control is not needed and the space voltage vector table is improved. The experimental results show that the torque ripple is reduced by 66.7%, the torque-ampere ratio is increased by 200% and the switching times in one electrical period are reduced by 47.3%.

This paper presents the wind tunnel experimental results to investigate the effects of surface gradient-of-curvature on aerodynamic performance of a low Reynolds number airfoil Eppler 387 for use in small-scale wind turbines. The prescribed surface curvature distribution blade design method is applied to the airfoil E387 to remove the gradient-of-curvature discontinuities and the redesigned airfoil is denoted as A7. Both airfoils are manufactured with high precision to reflect the design. Low-speed wind tunnel experiments are conducted to both airfoils at chord based Reynolds numbers 100 000, 200 000, and 300 000. Surface pressure measurements are used to calculate the lift and pitching-moment data, and the wake survey method is applied to obtain the drag data. The experimental results of E387 are compared with NASA Low Turbulence Pressure Tunnel (LTPT) results for validation. The gradient-of-curvature discontinuities of E387 result in a larger laminar separation bubble which causes higher drag at lower angles of attack. As the angle of attack increases the separation bubble of the airfoil E387 moves faster towards the leading edge than that of A7, resulting in a premature bubble bursting and earlier stall on E387. The impact of the gradient-of-curvature distribution on the airfoil performance is more profound at higher angles of attack and lower Reynolds number. The aerodynamic improvements are integrated over the 3D geometry of a 3 kW small wind turbine, resulting in up to 10% increase in instantaneous power and 1.6% increase in annual energy production. It is experimentally concluded that an improved curvature distribution results in a better airfoil performance, leading to higher energy output efficiency.

An efficient improved discrete velocity method (IDVM) with inner iteration is presented to simulate the steady flows in all flow regimes in this work. It is an extension of our previous implicit IDVM to achieve a faster convergence rate. In the previous method, both the discrete velocity Boltzmann equation (DVBE) and the corresponding macroscopic governing equations are solved synchronously, where the computational discrete cost is dominated by the calculation of the DVBE since the number of distribution functions is far larger than that of macroscopic conservative variables. Furthermore, the convergence rate of the calculation of the DVBE is affected by the predicted equilibrium state obtained from the solution of macroscopic governing equations. To provide a more accurate predicted equilibrium state for the fully implicit discretization of the DVBE, an inner iteration is introduced into the solution of macroscopic governing equations, and the flux Jacobian of these equations is evaluated by the difference of numerical fluxes of Navier–Stokes equations rather than the Euler equation-based flux splitting method used in the previous implicit IDVM. This more accurate prediction procedure endows the developed method to accelerate the computation greatly, especially in the continuum flow regime. Numerical results indicate that, in the continuum flow regime, the present method is about one order of magnitude faster than the previous implicit IDVM and one to two orders of magnitude faster than the conventional semi-implicit DVM.