Modelling of premixed turbulent propagating flames
Combustion has an active role in our modern lives as we continue to exploit its potential for many of our requirements. For example, its use to produce electricity and to power land, air and space transport vehicles. Increasing competition from the onset of the Industrial Revolution has led to a greater emphasis on improving technology. Furthermore, the ongoing issue of global warming has led to government legislation on emissions. These problems have led to increasing interest in gaining fundamental critical details on flow and combustion in simple and complex engineering geometries. Over the past twenty to thirty years numerical methods have demonstrated their success at obtaining information on flow and combustion. However, there is a continuing need to develop many of the components comprising a numerical method. The work reported here stems from the modelling of turbulent premixed flames. Turbulent premixed flames is a mode of combustion where the fuel and air mix before reacting. Such a combustion mode is present in spark ignition (SI) and gas turbine (GT) engines, and in explosions. Modelling of the combustion process within these practical applications can provide useful information. For example, in aiding the design of the piston bowl and the combustion chamber of SI and GT engines, respectively. Furthermore, the simulation of explosions can result in safer designs for fuel storage and supply facilities. A central parameter to be modelled in turbulent premixed flame propagation is the rate of chemical reaction. This is a crucial parameter since it controls the rate of flame propagation, flame structure, and resulting pressure history. However, to date the challenge of accurately modelling the rate of chemical reaction over a range of turbulence conditions remains. Therefore, in this thesis, mathematical models for the mean rate of reaction are examined, developed, and validated against time-resolved experimental data. The aim of the work is to improve the modelling of the mean rate of reaction in order to achieve closer agreement with available experimental results on rates of flame propagation, flame structure, and pressure history. Recent, practical and numerical experiments have provided support for algebraic and transport equation models for the flame surface area to volume ratio to model the mean rate of reaction. Here, these models are examined and developed with one-, two-, and three-dimensional computational fluid dynamics (CFD) calculations. The simulations were carried out using both an in-house code (Turbulent Reacting Flows, TRF) and a commercially available CFD code (FIRE). The TRF code was used to investigate the ability of existing and developed models to accurately predict turbulent burning velocity. The models were then validated further by simulating turbulent flame propagation in two combustion chamber configurations with built-in solid obstacles. Hence verifying the models for different turbulence and geometry conditions. A nonlinear eddy-viscosity model was implemented into the TRF code to assess the significance of turbulence modelling in turbulent premixed flames. Finally, the developed models were implemented in the FIRE code to carry out three-dimensional calculations to verify reproducibility of the TRF code results and to investigate secondary flow effects. Two reaction rate models were developed namely the algebraic (BML) and transport flame surface density (FSD) models. Both BML and FSD models yield plausible results for flame propagation in turbulent premixed combustion. However, modifications to the BML model were required for low turbulence conditions, and superior results were obtained with the FSD model. Both models struggled in capturing the interaction between flame and turbulent wakes behind obstacles when the standard linear eddy-viscosity turbulence was used. However, the application of a non-linear version of the eddy-viscosity model yields improved results for flame structure and speed around the obstacle, highlighting the importance of the turbulence model. The 3D calculations using the developed combustion model show good reproducibility of the 2D findings. Furthermore, the flame propagation, pressure history, and flame speed results are found to be in plausible agreement with the experimental data. It is shown that secondary flow mainly has the effect of increasing the rate of flame propagation in the single obstacle combustion chamber, and that the influence of secondary flow is dominant in the turbulent wake behind the obstacles.