A novel procedure for the topology optimization of an engine exhaust mixer

Abstract

The design of the engine primary exhaust of a helicopter must guarantee that the backpressure induced at the turbine outlet is sufficiently low to expel the hot air from the vehicle without compromising engine performances. At the same time, a large enough mass flow of air from the engine bay (secondary flow) is imposed to properly ventilate the bay. To this purpose, the exhaust is composed of two parts: the primary exhaust which conveys the air flow from the engine outlet and the secondary flow which collects the air from the primary exhaust and the air from the engine bay. The exhaust is designed with the aim of minimizing the backpressure at the engine outlet (minimizing therefore the fuel consumption) and of providing a secondary flow to properly ventilate the engine bay. In this paper, the discrete adjoint of the compressible Navier-Stokes equations coupled with the level-set method for the topology optimization is applied to this complex aerodynamic scenario, with the aim of designing a geometry able to minimize the impact on engine fuel consumption while fulfilling all other design requirements. The design case analyzed is the primary exhaust of the tilt-rotor demonstrator developed by Leonardo Helicopter division within the Clean Sky 2 Fast RotorCraft framework. Since the topologic optimization has been only recently introduced in Computational Fluid Dynamics tool, this work duly investigate all the steps required for the optimization. These include the choice of the constraints and the analysis of the error introduced by the Brinkman Penalization Model, which is used to simulate the solid domain introduced in the fluid field including some detail on modelling the walls generated by the optimization process. The impacts of the developed solutions are at the end compared with the primary exhaust developed using standard methodologies. Two novel geometries are produced which fulfill all design requirements. All simulations were performed using Siemens PLM commercial software STAR-CCM+.