The technique of pre-swirling cooling air to reduce its relative total temperature, as felt by rotating components, is well established. It is important to optimise the design of such systems in order to achieve maximum cooling effectiveness and to minimise the impact on cycle efficiency. Traditionally, these cooling systems have been developed by a combination of experimental investigation and careful evolution. However, more recently it has become practical to apply CFD to such problems. The nature of gas turbine cooling systems generally mandates the presence of discrete features on both static and rotating components, so that a fully rigorous analysis would need to be both 3D and unsteady, with the sub-domains adjacent to static and rotating surfaces solved in an appropriate frame of reference, together with a suitable interfacing procedure to communicate the evolving solution between each sub-domain. Such analyses are challenging for current CFD codes, both in terms of computation time and numerical stability. The present work explores the various options that are available to make such computations more practical and hence more accessible to the secondary systems modelling community. Significant reductions in set-up time can be achieved by adopting unstructured calculational meshes, although this may be at the expense of some loss of accuracy and increase in computational time relative to structured meshes. In the present work, an attempt has been made to quantify the effect of these choices. Depending on the configuration of the system under investigation, it may be permissible to ignore the unsteady interactions and to model the system using the more computationally efficient multiple reference frame (MRF) approach. Guidelines are proposed for assessing the likely impact of these simplifications on the results obtained.

This content is only available via PDF.
You do not currently have access to this content.