As CFD has matured to the point that it is capable of reliable and accurate flow simulation, attention is now firmly fixed on how best to deploy that CFD as part of a process to improve actual products. This “process” consists of capturing and controlling the geometry of a suitable portion of an aeroengine (e.g., a blade row, or an internal cooling system or a fan-plus-nacelle), building a mesh system, solving the flow and responding to an appropriately visualized flow field by changing or accepting the geometry. This paper looks at that process from the point of view of identifying any bottlenecks and argues that current research should be directed at the CAD-to-mesh-to-solution cycle time rather than, as has been traditional, just looking at the solver itself and in isolation. Work aimed at eliminating some of these bottlenecks is described, with a number of practical examples.

As CFD has matured to the point that it is capable of reliable and accurate flow simulation, attention is now firmly fixed on how best to deploy that CFD as part of a process to improve actual products. This “process” consists of capturing and controlling the geometry of a suitable portion of an aeroengine (eg a blade row, or an internal cooling system or a fan-plus-nacelle), building a mesh system, solving the flow and responding to an appropriately visualized flow field by changing or accepting the geometry. This paper looks at that process from the point of view of identifying any bottlenecks and argues that current research should be directed at the CAD-to-mesh-to-solution cycle time rather than, as has been traditional, just looking at the solver itself and in isolation. Work aimed at eliminating some of these bottlenecks is described, with a number of practical examples.

The introduction of lean premixed combustion technology in industrial gas turbines has led to a number of interesting technical issues. Lean premixed combustors are especially prone to acoustically-coupled combustion oscillations as well as to other problems of flame stability such as flashback. Clearly it is very important to understand the physics that lies behind such behaviour in order to produce robust and comprehensive remedies, and also to underpin the future development of new combustor designs. Simulation of the flow and combustion using Computational Fluid Dynamics (CFD) offers an attractive way forward, provided that the modelling of turbulence and combustion is adequate and that the technique is applicable to real industrial combustor geometries. The paper presents a series of CFD simulations of the Rolls-Royce Trent industrial combustor carried out using the McNEWT unstructured code. The entire combustion chamber geometry is represented including the premixing ducts, the fuel injectors and the discharge nozzle. A modified k-ε model has been used together with an advanced laminar flamelet combustion model that is sensitive to variations in fuel-air mixture stoichiometry. Detailed results have been obtained for the non-reacting flow field, for the mixing of fuel and air and for the combustion process itself at a number of different operating conditions. The study has provided a great deal of useful information on the operation of the combustor and has demonstrated the value of CFD-based combustion analysis in an industrial context.

No abstract

An assessment of various automatic block topology generation techniques for creating structured meshes has been performed in the first part of the paper. The objective is to find out optimal blocking methods for generating meshes suitable for flow simulations. The comparison has been carried out using an adjoint based error analysis of the meshes generated by these block topologies. Different objective functions and numerical schemes have been used for this assessment. It is found that, in general, the medial axis based approaches provide optimal blocking and yields better accuracy in computing the functional of interest. This is because the medial axis based methods produce meshes which have better flow alignment specially in case of internal flows. In the second part of the paper, the adjoint based error indicator has been used to adapt the block topology in the regions of large error.

No abstract

The modern CFD process consists of mesh generation, flow solving and post-processing integrated into an automated workflow. During the last several years we have developed and published research aimed at producing a meshing and geometry editing system, implemented in an end-to-end parallel, scalable manner and capable of automatic handling of large scale, real world applications. The particular focus of this paper is the associated unstructured mesh RANS flow solver and the porting of it to GPU architectures. After briefly describing the solver itself, the special issues associated with porting codes using unstructured data structures are discussedfollowed by some application examples.

No abstract