A Fully Integrated Approach to Component Zooming Using Computational Fluid Dynamics

Background. This study focuses on a simulation strategy that will allow the performance characteristics of an isolated gas turbine engine component, resolved from a detailed, high-fidelity analysis, to be transferred to an engine system analysis carried out at a lower level of resolution. This work will enable component-level, complex physical processes to be captured and analyzed in the context of the whole engine performance, at an affordable computing resource and time. Approach. The technique described in this paper utilizes an object-oriented, zero-dimensional (0D) gas turbine modeling and performance simulation system and a high-fidelity, three-dimensional (3D) computational fluid dynamics (CFD) component model. The work investigates relative changes in the simulated engine performance after coupling the 3D CFD component to the 0D engine analysis system. For the purposes of this preliminary investigation, the high-fidelity component communicates with the lower fidelity cycle via an iterative, semi-manual process for the determination of the correct operating point. This technique has the potential to become fully automated, can be applied to all engine components, and does not involve the generation of a component characteristic map. Results. This paper demonstrates the potentials of the “fully integrated” approach to component zooming by using a 3D CFD intake model of a high bypass ratio turbofan as a case study. The CFD model is based on the geometry of the intake of the CFM56-5B2 engine. The high-fidelity model can fully define the characteristic of the intake at several operating condition and is subsequently used in the 0D cycle analysis to provide a more accurate, physics-based estimate of intake performance (i.e., pressure recovery) and hence, engine performance, replacing the default, empirical values. A detailed comparison between the baseline engine performance (empirical pressure recovery) and the engine performance obtained after using the coupled, high-fidelity component is presented in this paper. The analysis carried out by this study demonstrates relative changes in the simulated engine performance larger than 1%. Conclusions. This investigation proves the value of the simulation strategy followed in this paper and completely justifies (i) the extra computational effort required for a more automatic link between the high-fidelity component and the 0D cycle, and (ii) the extra time and effort that is usually required to create and run a 3D CFD engine component, especially in those cases where more accurate, high-fidelity engine performance simulation is required.