As aircraft manufacturers push for better fuel economy, novel aircraft concepts, like the propulsive fuselage concept have emerged. One of the prime component of such concepts is the boundary layer ingestion (BLI) engine. Such engine operates by ingesting the low-momentum fluid e
...
As aircraft manufacturers push for better fuel economy, novel aircraft concepts, like the propulsive fuselage concept have emerged. One of the prime component of such concepts is the boundary layer ingestion (BLI) engine. Such engine operates by ingesting the low-momentum fluid emerging from the surface of the aircraft fuselage which enables to enhance its propulsive efficiency. The design of a BLI engine is fundamentally different than that of a conventional aircraft engine as the BLI fan experiences distorted airflow at every revolution. Therefore, when its performance are analyzed by means of CFD, single passage simulations become insufficient to represent the flow characteristics. and extremely expensive simulation of the entire annulus become necessary for analysis and design purposes. Especially during conceptual design, where several BLI engine configurations must be investigated in combination with their airframe integration, new reduced-order methods are required to alleviate the computational burden while retaining a sufficient level of accuracy.
An alternative, computationally efficient approach capable of coping with the full-annulus simulation is the well-established body-force modelling (BFM). BFM is a technique through which the physical blade rows are replaced by a volumetric force field which mimics the flow turning imparted by the physical blade rows. Thanks to this simplification, BFM does not require the physical blades to be accurately reproduced in the mesh, thus offering substantial computational cost reduction. Due to the associated gains in computational efficiency, this method is commonly used for the analysis and off-design performance prediction of full-annulus geometries, which are computationally demanding through conventional methods. Although the method has been established for analysis purposes, its capability as a design tool was never assessed.
Implementation of BFM in a CFD code that can be automatically differentiated to attain design sensitivities via the adjoint method may enable efficient BFM-based design of complex engine configurations.
In this research, the BFM model originally developed by Ref. \cite{Hall2015AnalysisIngestion} and improved by Ref. \cite{Thollet2017BodyInteractions} was implemented into open-source CFD software SU2, which is equipped with a discrete adjoint solver based on operator overloading. In addition, parallel force and a metal blockage source term were added to the existing formulation to increase the fidelity of the BFM. The implementation was validated by comparing the results obtained from the BFM and the physical blade computation on an exemplary axial turbine test case.
The results show that the BFM was capable of providing static and stagnation pressure and temperature trends with an accuracy of approximately 94\%. Furthermore, the absolute flow deflection by the stator and rotor rows were found to deviate by $\SI{5}{\degree}$ and $\SI{17}{\degree}$. The BFM was found to be 3 orders of magnitude faster than the equivalent physical blade computation.