Gas turbine engine starting models require a lot of calibration to represent reality with acceptable accuracy due to the lack of high-quality component rig data in the sub-idle region. A detailed sensitivity study is presented in this paper to guide such calibration efforts. A thermodynamic component-matching type transient model of a single-spool turbojet engine with shaft and heat-soakage dynamics is employed for this purpose. Turbomachinery component maps are extended to sub-idle using an in-house map smoothing tool and the strategies presented by Kurzke recently. These extension strategies make use of the correlations hidden in the already available regions of the maps and ensure physical consistency. However, they contain some uncertainty, even when an experimentally obtained zero-speed line is available. Combustor sub-idle efficiency, stability limits, and delay are taken from the literature. Due to the chaotic nature of a combustor in the sub-idle region, a precise prediction of the combustor efficiency seems impossible. Effects of uncertainties related to sub-idle turbomachinery map extensions, burner efficiency, and heat soakage are investigated in this paper. Two popular fuel control strategies are employed and compared to see how controls deal with these uncertainties. It is concluded that turbomachinery torque characteristics and turbine capacities are the most important parameters when calibrating a starting model with a control based on rotational acceleration while burner efficiency and heat soakage are added on top of these with a control based on fuel flow rate.

This study presents the first results of a new turbulence model implementation in our compressible finite volume CFD solver. The k - kL turbulence model is one of the newest two-equation models, and it is based on the ideas of Rotta’s two-equation model. Various research groups progressively develop the model, and it is maturing rapidly. Reports suggest that the k - kL turbulence model provides superior results compared to the other two-equation turbulence models in specific problems. The improved solutions are observed mainly for the flows with high adverse pressure gradients, the blunt-body wakes and jet interactions. We have implemented the k - kL model (with the standard designation of k-kL-MEAH2015) in our solver, and we are testing it rigorously. This paper presents our results on standard turbulence test cases: subsonic flat plate and subsonic wall-mounted bump. The results compare well with the reference study previously presented and published by model developers. The design of the k - kL model prevents excessive production of turbulence and dissipation; hence it preserves vortices significantly better than the other two-equation models. The implemented model is also tested with a transonic fin trailing vortex case to support this statement. Results show that the k-kL model yields considerably better results than the SST turbulence model in cases including vortices.

In this study, we present our improved RANS results of the missile aerodynamic flow computation involving leading edge vortex separation. We have used our in-house tailored version of the open source finite volume solver FlowPsi. An ongoing study in the NATO STO Applied Vehicle Technologies Panel (AVT-316) has revealed that a highly maneuverable missile configuration (LK6E2) shows unusual rolling moment characteristics due to the vortex–surface interactions occurring during wing leading edge separation of vortices. We show the performance of the recently developed k-$kL$ turbulence model for this test problem. This turbulence model is shown to have superior capabilities compared to other widely used turbulence models, such as Spalart–Allmaras and shear stress transport. With the k-$kL$ turbulence model, it is possible to achieve more realistic computational results that agree better with the physical data. In addition, we propose improvements to this turbulence model to achieve even better predictions of rolling moment behavior. Modifications based on turbulence production terms in the k-$kL$ turbulence model significantly improved the predicted rolling moment coefficient, in terms of accuracy and uncertainty.