Computation of radar cross section of jet engine inlets with a nonuniform cross section and complex internal structures

摘要

Electromagnetic scattering from jet engine inlets is an important topic of research due to the significant contribution that the inlet has to the radar cross section (RCS) of modern aircraft. Many researchers have studied this topic using an assortment of different methods. Some techniques applied to the problem are standard methods such as mode matching (MM), shooting and bouncing ray (SBR), as well as finite element method (FEM) and method of moment (MoM) techniques. However, it is well known that these standard methods all have some serious drawbacks that prevent them from being used to analyze and quantitatively design practical jet engine inlets. In the case of MM, the resulting system demonstrates slow convergence and requires a large number of modes as well as a poorly conditioned large matrix to invert. Furthermore, rigorous modal methods are only useful for simple geometries that can have canonical terminations. The problems with SBR techniques are that while they are simple and efficient, they lack the required accuracy to handle arbitrary complex structures and material compositions. Standard FEM and MoM have trouble handling the entire structure due to their large computational requirements. More recently, hybrid methods have been introduced to solve this important problem. Some of them combine high-frequency techniques in the empty part of the inlet with numerical methods in the loaded section. Despite the advantages of these methods they still require a large number of degrees of freedom which makes them very difficult to apply to large structures. A particular method that is very well suited to the computation of RCS of jet engine inlets is the finite element-boundary integral (FE-El) method [1,2]. Since jet engine inlets are complex geometries that may have complex material composition, this is well suited for a finite element approach. The domain size is truncated with a boundary integral, which reduces the overall simulation domain. Furthermore, the techniques described in [1] and [2] exploit the structure of the FE-El equations to require minimal memory (only proportional to the maximum cross section of the inlet and independent of its depth) and computation time which increases linearly with respect to the inlet depth.