Propeller Analysis with Adjoint Method using SU2
More Info
expand_more
Abstract
To cater to the increasing demand in short-range passenger air transport and the research in electric and hybrid aircraft, there is a renewed interest in the research of propellers. Propellers are known to have high aerodynamic efficiency, which translates to lower fuel burn in comparison to other jet engines, making it an attractive option for airlines to operate it in regional transport. With such a heightened interest in the research of propellers for future applications, it is crucial to study the effect of the shape-design of the blades on its aerodynamic performance parameters. The combination of advanced numerical methods with computational fluid dynamics allows determining the sensitivity of cost function to variations of the design parameters. In this thesis, a RANS-based CFD methodology incorporating the adjoint method is developed to perform the sensitivity analysis of the propeller's aerodynamic performance to the variation in the shape design of the blade. This methodology is applied to two propeller test cases- straight and swept blade. The propeller aerodynamic analysis is performed with SU2 for a range of advance ratios and compared against ANSYS Fluent's. Based on the verification and validation of SU2's flow solver, a larger discretization error was obtained with SU2 in comparison with ANSYS Fluent, for the simulations on the same computational mesh. Once SU2's flow solver is verified and validated using the two propeller cases, it is used to draw a comparison between the aerodynamic performance of the two blades. the swept blade operated with higher aerodynamic efficiency at high advance ratios with the largest gain of 0.47% at J=1.2. While it experiences a marginal loss in the efficiency at lower advance ratios with the largest decrement being 0.37% at J=0.6.
SU2’s adjoint solver is used to perform discrete adjoint-based sensitivity analysis of the propellers at multiple design points with thrust coefficient and torque coefficient as the objective functions. The sensitivity analysis of the thrust coefficient suggested that the regions of high sensitivity are located along the leading-edge and outboard portions of the blade's suction side. The concentration of the regions of high sensitivity varied with the advance ratio. For a given blade, similar trends were seen for the surface sensitivities of the torque coefficient. Both blades exhibit similar trends for surface sensitivities of both objective functions. Whereas based on the gradient information, the highest gradients of thrust coefficient are attained for the design variables located near the trailing edge of the straight blade, at a given radial position. Swept blade shows a chordwise gradient trend similar to the straight blade at most radial positions, except the tip region. It was inferred that the introduction of sweep at the tip results in the shift of the dominant design variable from trailing edge to mid-chord. Similar to surface sensitivities, trends of the chordwise gradient of two objective functions were alike for a given blade. Based on this study, the resulting understanding of the sensitivity of the propeller’s performance parameters to the variation in blade shape design can be used as an input for future design studies aimed at designing efficient propellers. Likewise, the RANS-based CFD methodology incorporating the adjoint method used in this work can be aimed at developing adjoint-based multidisciplinary design capabilities for propellers.