Design and Analysis of an Installed Pusher Propeller with Boundary Layer Inflow

Abstract

Boundary Layer Ingestion is an integrated propulsion concept in which a propulsor operates in boundary layer flow instead of the free streamflow with the goal to reduce the fuel flow for a given operating condition. The objective of this thesis is to obtain a better understanding of the power benefit of an installed pusher propeller at the aft fuselage by designing the aerodynamic shape of the propeller and validate the design by means of CFD simulations. A propeller analysis tool for uniform inflow (UI) and non-uniform inflow (NUI) named N-XROTOR is developed using the lifting line code XROTOR in combination with XFOIL to calculate airfoil properties. The tool is validated using experimental results and results from CFD simulations of uniform inflow propellers. N-XROTOR shows good agreement of the trend of the CT-J and CP-Jcurves but a constant over prediction of both thrust and power with respect to experimental data is observed and several deviations are explained. A series of CFD simulations in ANSYS Fluent™ using a reduced wedge shaped domain of one blade of the N250 propeller are performed for several advance ratios including a grid refinement study. Minor deviations between a transient and a steady simulation are found and the steady method is chosen based on computational cost. The trends of N-XROTOR in terms of CT-J and CP-J compare well with the CFD simulation with a constant over prediction of the performance quantities by N-XROTOR. These over predictions are also noticeable in the radial distributions of thrust and torque with slight over predictions in the high loaded region on the blade. For a moderate advance ratio of J = 0.79 the thrust and power are over predicted by 5.25% and 3.67% respectively. A comparison with the standard kapppa-omega SST turbulence model and the SST model with low Reynolds number correction is made. The radial flow on the propeller blade is shown to be quite significant and varies along the blade and shows good agreement with the distribution of bound circulation and the resulting trailing vorticity. A design procedure is developed in which the propeller shape is optimised using shape functions to describe the pitch and chord distribution and a NACA four series airfoil is used to limit the number of design variables for a gradient based optimisation algorithm in Matlab environment. The interaction effects are assumed to be determined a-priori and a tapered aft fuselage and the pressure field induced by the fuselage are neglected. Input quantities for the design routine include an inflow field from CFD analysis, the design advance ratio and a thrust requirement. The design objective of all optimisations is minimum power. For the reference design case an axisymmetric body from ESDU is subject to CFD simulations to obtain the inflow profile and fuselage drag for the isolated and installed configuration. Interference effects are approximated using an Actuator Disk (AD) model at the predefined location of the propeller with a pressure jump equal to the defect in total pressure in the boundary layer based on findings from previous research. An 11% increase of drag is found for the equilibrium condition which is primarily due to increased pressure drag. Larger pressure jumps show only a marginal increase in drag. In a comparison study, the number of blades is set to four, an advance ratio of J = 1.50 is chosen and in combination with a radius equal to 99% of the total gage pressure of the undisturbed air yields a tip Mach number of around 0.50. The optimisation results show that the NUI propeller requires 6.93% less power compared with the UI propeller despite the 11% higher thrust. The thrust distribution of the NUI propeller shows a significant increase in thrust in the lowaxial velocity region towards the root and the maximum thrust is shifted inboard. The ratio of thrust to power dT/(dQ Omega)­ along the propeller blade shows a constant distribution for the UI propeller, while the NUI propeller has a smooth increasing distribution towards the root. This distribution shows that thrust requires a relatively low power when the local axial velocity is relatively low. It is found that this is the main benefit of positioning a propeller in the boundary layer. The bound circulation distribution shows a shift towards the root compared with the distribution of the uniform inflow propeller which is the result of the optimised propeller shape which benefits from the favourable thrust to power ratio in the inner radii. The NUI propeller has a significant increased chord compared with the optimal UI and also a higher lift coefficient distribution. The local efficiency defined as eta = dTVa/(dQ Omega) with Va as the local inflow velocity. Optimal UI propellers have a constant efficiency distribution, but the NUI propeller shows a decreasing trend towards the root which is also found in literature. The trend of lower local efficiency is also found when an optimisation for minimum power is performed using a radially varying actuator disk with the same inflow and thrust requirement as for the full blade propeller. Additional analysis on the NUI propeller include a comparison of off design conditions and additional optimisations are performed to quantify the effect of the number of blades, radius and advance ratio. The optimised NUI propeller in the installed configuration is simulated using CFD. N-XROTOR over predicts the thrust and power by 4.15% and 4.71% respectively compared with the CFD simulation, which are deviations of the same order as the N250 simulation. The thrust to power distribution shows good correspondence. In the root region this ratio is under predicted by N-XROTOR which is expected to be the result of a large pressure and velocity gradient at the junction of the spinner and propeller surface resulting in a region of recirculation. Also the blockage effect of the tapered spinner results in larger angles of attack in the root region. The outer region shows trailing edge stall which is found to be primarily due to the coarse mesh in that region. Improved results are obtained when N-XROTOR uses airfoil data obtained from two-dimensional CFD analysis of a particular airfoil section. The kappa-omega SST model with low Reynolds number correction shows almost exact agreement with XFOIL. The standard turbulence model shows a decambering effect and an earlier stall behaviour. The remaining deviations between N-XROTOR with approximated CFD airfoil properties are expected to originate from the radial flow on the blade and the variation in circulation in chordwise directions which are not simulated in N-XROTOR. Both the externally induced radial flow by the tapered aft fuselage and the self induced radial flow are expected to result in a decambering of the airfoil due to the influence on boundary layer growth as well as a reduced chordwise velocity resulting in a locally lower dynamic pressure experienced by the airfoil contour. The interference effects of the propeller onto the fuselage are compared with the Actuator Disk (AD) approximation. An over prediction of 0.74% of the drag by the AD model of the fuselage excluding spinner is observed. Downstream of the full blade simulation the pressure is rapidly decreased to a low finite value at the aft end of the spinner. This is the result of the finite bound circulation at the propeller root which releases a strong trailing vortex from each blade. These vortices combine into a strong axial vortex which induces a strong tangential velocity and therefore in a low pressure acting on the spinner. A slipstream analysis is performed of circumferentially averaged flow quantities in radial direction at a plane behind the propeller and the axial development of several averaged flow quantities is shown. Several recommendations for future work are formulated to improve the propeller design, improve the design procedure, reduce the interference effects and increase the power benefit of the non-uniform inflow propeller.