Innovation in Engineering Computational Technology
Chapter
12 A.C. Benim*, A. Nahavandi* and K.J. Syed+
*Department of Mechanical and Process Engineering, Düsseldorf University of Applied Sciences, Germany +Combustion Engineering Department, Siemens Industrial Turbomachinery, Lincoln, United Kingdom Keywords: turbulent swirling flows, turbulence modelling, URANS, RSM, LES, DES.
Owing to their importance in a broad range of applications, swirling flows have been the subject of intensive experimental, analytical and numerical investigations over many years [1,2,3,4,5,6,7]. The application of swirling flow in industrial gas turbine combustors is of particular interest to the current work. Lean, premixed, dry, low emission combustors exhibit flows that can be substantially different from non-premixed ones, due to a significantly larger degree of swirl acquired by the flow. A vortex breakdown recirculation zone along the swirl axis can be seen as an abrupt transition between an upstream super-critical flow and a downstream sub-critical flow [8]. The criticality of the swirling flow describes whether the axial flow velocity exceeds the relative phase velocity of upstream-directed inertia waves (super-critical flow) or vice versa (sub-critical flow). The influence of the downstream conditions can hence be transmitted upstream where the flow is sub-critical, but only affects the local flow at the exit where it is super-critical. Although many analytical and numerical investigations have been conducted to address vortex breakdown and the criticality of the vortex core in confined turbulent flows [1,2,3,4,5,6,7] much of this work is based upon simplifying assumptions such as inviscid flow and axisymmetry. These studies have revealed much about the behaviour of vortical flows but they fall short of delivering quantitative information that can be used in the design of combustors. Only a CFD-based approach that deals with turbulence heat release and the complexities of the geometry can address the key details relevant to the state of the vortex core. CFD suffers, however, from numerical errors due to flow and geometry complexities and practical grid resolution and the need for physical models for combustion and turbulence. Thus, if CFD methods are to be used effectively in the design of swirl combustor, validation is required to understand the impact of the numerical and modelling deficiencies. This is the scope of the present study. The present work aims to provide a detailed validation of a broad range of advanced turbulence modelling strategies for this bracket of flows problems. For validating the results, the experimental data of Escudier and Keller is [9] utilised, which is obtained at a water-test rig, for flows exhibiting similar characteristics to those typically encountered in gas turbine combustors. The analysis is based on the commercial general-purpose CFD code ANSYS-CFX [10]. For modelling the turbulence, the Reynolds stress model-based (RSM) [11], unsteady Reynolds averaged numerical simulations (URANS), large eddy simulations (LES) [12] and detached eddy simulations (DES) [13] approaches are adopted. It is observed that better results are obtained using the RSM based URANS approach. DES is also observed to perform better than LES. The rather unsatisfactory performance of LES may be attributed to inaccuracies in formulating the boundary conditions, and/or because there was too coarse a cut-off scale for the simple sub-grid model adopted. Although the present work shows that remarkable improvements can be achieved compared with traditional RANS-based approaches, comparisons between the predictions and the measurements indicate that further improvements are necessary in order to achieve a satisfactory predictive capability for turbulent swirling flows.
References
|