The use of 3D CFD combustion models based on tabulated chemistry is becoming increasingly popular. Especially the runtime benefit is attractive, as the tabulated chemistry method allows to include state-of-the-art chemical reaction schemes in CFD simulations. In this work, the Tabkin FGM combustion model in AVL FIRE™ is used to assess the predictivity on a large database of a light-duty Diesel engine measurements. The AVL TABKIN™ software is used to create the chemistry look-up tables for the Tabkin FGM model.
The TABKIN software has been extended with the kinetic soot model, where the soot mass fraction calculation is done during the chemistry tabulation process, as well as an NO model using a second progress variable. From recent validation studies, a best-practice and nearly automated workflow has been derived to create the look-up tables for Diesel engine applications based on minimal input. This automated modeling workflow is assessed in the present study.
A wide range of parameter variations are investigated for 5 engine load points, with and without EGR, in total 186 cases. This large number of CFD simulations is run in an automated way and the parameters of the CFD sub-models are kept equal as well as all numerical settings.
Results are presented for combustion and emissions (NO and soot). Combustion parameters and NO emissions correlate very well to the experimental database with R2 values above 0.95. Soot predictions give order-of-magnitude agreement for most of the cases; the trend however is not always respected, which limits the overall correlation for all cases together, as reported by other authors. Further fundamental research on modeling soot formation and oxidation process remains required to improve the models. In terms of CPU time, the present study was executed on an off-the-shelf HPC cluster, using 8 CPU cores per case and requiring around 3 hrs of wall-time per case, e.g. such a large set of calculations can be simulated overnight on a standard HPC cluster.