No abstract

An overview over flamelet modelling for turbuunder diesel engine conditions using n-heptane as lent non-premixed combustion is given. A short review of fuel. In 1996 Pitsch et al. [2] investigated the pollutprevious contributions to simulations of direct injection ant formation in a Volkswagen direct injection (DI) (DI) diesel engine combustion using the representative diesel engine, also fuelled with n-heptane. Different interactive flamelet concept is presented. A surrogate fuel exhaust gas recirculation (EGR) rates were conconsisting of 70 per cent (liquid volume) n-decane and 30 sidered and the well-known trade-off between NO x per cent a-methyl-naphthalene is experimentally comand soot was reproduced by the simulations. The pared to real diesel fuel. The resemblance of their physical significance of the scalar dissipation rate repand chemical properties is shown to result in very similar resenting strain effects on the ignition and combuscombustion and pollutant formation for both fuels. tion process, especially the ignition delay time, was In order to account for variations of the scalar dissishown. pation rate within the computational domain, a method Barths et al. [3] compared results of simulations using multiple flamelets, called the Eulerian particle using an eddy dissipation model [4] in combination flamelet model, is used. A strategy is described for subwith the extended Zel'dovich NO x mechanism and dividing the computational domain and assigning the the Hiroyasu soot model [5] (see the Appendix) to resulting subdomains to different flamelet histories repthose obtained with the RIF model. The tuning parresented by Eulerian marker particles. Experiments conameters (for ignition and combustion) in the former ducted with an Audi DI diesel engine and diesel fuel are models were adjusted for the baseline case (0 per compared to simulations using the surrogate fuel. The use cent EGR) and then held constant. The constants for of multiple flamelets, each having a different history, sigthe NO x and soot models were set as recommended nificantly improves the description of the ignition phase, by Belardini et al. [6]. The trade-off curves for both leading to a better prediction of pressure, heat release and models and the experiments are shown in Fig. 1. exhaust emissions of soot and NO x. The effect of the The RIF model predicts the NO x emissions very number of flamelet particles on the predictions is well and slightly over-predicts the soot emissions, discussed. which, however, are still in the correct order of magnitude. The trends for NO x and soot are also well Key words: CFD modelling, diesel engines, combustion, reproduced. The Magnussen model [4] shows a sigflamelet, emissions nificantly stronger decrease of the NO x emissions on the EGR rate than the experiment. The extended Zel'dovic mechanism only accounts for thermal NO,

Large eddy simulations of pulverised coal combustion (PCC-LES) stabilised on a laboratory-scale piloted jet burner are carried out. The joint simulation effort of three research groups at Freiberg University (FG), Imperial College (IC) and Stuttgart University (ST) is presented, and the details of the comprehensive coal combustion models and their numerical implementation in three different computer programs are discussed. The (standard) coal sub-models and parameters used by the different groups are unified wherever possible. Differences amongst the groups are a different code basis and an Eulerian treatment of the coal particles by IC, while FG and ST use the Lagrangian framework for particle transport. The flow modelling is first validated for the corresponding non-reacting case and all LES calculations accurately capture the experimental trends. Velocity field statistics for the PCC case are in good accordance with the experimental evidence, but scalar statistics illustrate the complexity of coal combustion modelling. The results show notable differences amongst the groups that cannot only be attributed to the different treatment of the particle phase, and they highlight the difficulty to assess and interpret the quality of specific modelling approaches, and a need for further work by the research community. The present study is the first to compare three originally independent transient coal simulations and a step towards comprehensive PCC-LES

No abstract

A detailed multi-purpose reaction mechanism for ethanol combustion was developed for the use in high-fidelity numerical simulations describing ignition, flame propagation and species concentration profiles with high accuracy. Justified by prior analysis, an optimization of 44 Arrhenius parameters of 14 crucial elementary reactions using several thousand direct and indirect measurement data points was performed, starting from the ethanol combustion mechanism of Saxena and Williams (2007). The final optimized mechanism was compared to 13 reaction mechanisms frequently used in ethanol combustion with respect to their accuracy in reproducing the various types of experimental data.