Modelling and design of LPC components with semi-analytical models Part II: design of optimal OGV cascades

The growth in diameter of turbofan engines exacerbates problems related to the interaction of the Outlet Guide Vanes (OGV), pylon and intake because it reduces the ratio between components gaps and disturbance wavelength. The main components of this interaction are the potential fields generated by the intake and by structural components in the bypass, the pylon and the Radial Drive Fairing (RDF). The OGV bladerow and the fan are immersed in these potential fields and suffer performance degradation as well as integrity issues as a result. Simple actuator-disc analysis shows that a uniform OGV cascade amplifies the effect of the pylon potential flow. Therefore, a number of methods have been proposed over the years to compute OGV exit flow angle patterns that result in an approximately circumferentially uniform static pressure field at fan exit. Within actuator disc approximations, the determination of the optimal exit flow angle pattern can be accomplished analytically but little information is obtained on how the geometry of the vanes ought to be modified. Consequently, it is not difficult to generate by this method OGV cascades that stall or choke locally. More recent contributions use CFD computations coupled to optimization methods to determine OGV patterns that reduce the distortion at the fan exit, while minimising some measure of OGV loss. Whilst in principle more rational, these methods encounter practical difficulties due the computational power needed to obtain reliable loss estimates while exploring large design spaces. In this paper the view is taken that the performance of the OGV bladerow can be preserved during the optimization process if the loading distribution of each vane is made to match the loading distribution of the nominal vane (i.e. the aerodynamic design intent with axisymmetric inlet and exit flow). As loading distributions are readily available from inviscid-type analysis, the generation of optimal OGV patterns can be accomplished with very reasonable computational expense using a method based on the model described in part I of this paper.