OCR Text |
Show Abstract Results for the swirling, un-staged, 300 kW BERL natural gas flame have been obtained using the CFD code FLUENT fUNS. The solver is based on a fully unstructured finite volume scheme. The predictions have been compared to those obtained using the structured solver FLUENT. Some comparisions with measurements are also made. 'IUrbulence is modeled by means of the standard k - f model. Two different models to describe the turbulence chemistry interaction have been considered : i) a conserved scalar pdf model, ii) the Magnussen-Hjertager eddy breakup model. For the first sub-model, the f3 probability density function is used, while the second uses eit her a single step reaction scheme or a two-step reaction scheme. Radiation is computed using the P-l differential approximation and the WSGGM (Weighted Sum of Gray Gases Model) is used to calculate the radiative properties. Comparison of velocities, temperature and species concentrations for both FLUENT and FLUENT/UNS using the conserved scalar pdf model have been made at three stations downstream of the quarl. Predictions of velocities, t emperature and species using the different schemes (pdf, Magnussen single step, and Magnussen two-step) are also compared to experimental data at some stations. Introduction Numerical modeling of natural gas combustion is a complex phenomenon influenced by a large number of interacting processes. Comprehensive testing and validation is necessary to establish the range of applicability and level of confidence in commercial CFD codes that can eventually be used for design of industrial burners. This task is often made difficult by the lack of data from suitable benchmark industrial problems. The present work resulted from validation studies carried out at Fluent Inc. for the experiments conducted at the Burner Engineering Research Laborat.ory (BERL) as part of the "Scaling 400" study, whereby experimental measurements were taken for combust.ors ranging in size from 300 kW to 12 MW. This particular study deals with an unstaged 300 kW, nat.ural gas flame. Data from the experiments were collected by Sayre et al. and are published in [1].The quality of t.he present. benchmark[l] is deemed to be excellent insofar as the geometry and flow boundary conditions are well defined . Besides, the extensive measurements are well suited for code validation. Objectives FLUENT/UNS is a general purpose CFD code that was released in June 1996. It offers several advantages over conyent.ional codes t.hat are based on a structured framework, for instance, decreasing time used in setting up the geometry and grid, allowing adaption to evolving solutions to capture high gradients, true dynamic allocation of memory and parallel processing. This paper is part of a two-step internal project: (i) Compare FLUENT/UNS predictions wit.h experimental data and previously obtained predictions for the 2-d axisymmetric case using FL UENT (ii) Compare predictions obtained from full 3-d simulations against experimental data. This paper reports t.he progress made toward the first part of the project. The principal purpose of this study was to validate predictions obtained by using a new unstructured solver, FLUENT/UNS, against predictions made by a structured solver, FLUENT. This was to ensure that the new solver could be used for similar applications with the same reliability that characterized the structured solver. Further, some comparisons were also made with appropriate in-flame measurements at various stations in the furnace. The objective is not to describe any new model of turbulence-chemistry interaction or combustion , but to validate existing models available in the literature and implemented in FLUENT/UNS. Problem Description The flame considered is unstaged natural gas fired in a 300 k W swirl-stabilized burner (Figure 1). The furnace is vertically fired and of octagonal cross-section, with a conical furnace hood and cylindrical exhaust duct. The furnace wall may either be lined with refractory material or water-cooled. The burner features 24 radial fuel 1 |