Within the current combustor design process, the combustor performance and sizing is mainly estimated using Prelim Design Tools, which are based on empirical correlations. In order to investigate these preliminary designs in more detail, the application of CFD within the preliminary combustor design process has steadily increased. However, the generation of CFD solutions is a time consuming process, since it requires adapted CAD models and the generation of meshes before the actual CFD computations can be performed. Moreover, it has to be assured that the meshes are suitable for the computational code, which may require several iterations within the grid generation process. Next the manual setting of the boundary conditions for CFD is also prone to errors and the post-processing of the CFD results is generally time consuming. Therefore, an advanced approach of an automated preliminary combustor design process has been developed within the 7th framework of the European research project IMPACT-AE, based on the knowledge obtained in the 6th framework of the European research project INTELLECT D.M. The processes developed are based on improved and robust knowledge-based-engineering tools for modern low NOx combustor geometries.
The advanced approach comprises a KBE preliminary design tool with improved Rolls-Royce in-house design rules for modern low NOx high efficiency combustors. It is based on fully featured 3-D parametric combustor models, including the pre-diffuser, the injector, the flame tube and the casing parts. An unstructured automatable meshing-tool CFS BOXERMesh is used to generate high quality meshes. Subsequent CFD analyses are performed using an unstructured Rolls-Royce in-house combustion CFD code PRECISE-UNS and the post-processing is performed using the open source tool ParaView. The choice of these tools is based on their ability to be fully automated using appropriate scripts and information. To create a strong linkage between these tools a KBE interface-tool and a database were implemented. These tools manage the individual processes fully automatically, including e.g. the transfer of geometry parameters to the CAD models; the gathering of the combustion chamber dependent boundary conditions and numerical parameters for the CFD analysis; or the generation of tool-internal automation scripts, using interface-tool input and database information.
Several grid refinement studies were performed to define an automated mesh refinement strategy, using scaled cell sizes for each combustor feature. Furthermore, a mixing port diameter scaling and a detailed post-processing procedure are described.
A Knowledge-Based Geometry Repair System for Robust Parametric CAD Models
