PhD thesis, Cranfield University, Bedford, UK, 1995
The thesis addresses the problem of predicting the emission of exhaust smoke from the diesel engine. A simulation program based on a zonal phenomenological combustion model was developed, permitting analysis of soot modelling techniques. For the first time, a comparative study of the common soot model expressions was undertaken. Model sensitivities and behaviour were critically assessed in order determine the key model parameters and to establish a more solid predictive capability. Validation of both the combustion and soot predictions was made by means of comparison with the extensive experimental data-set of Kamimoto, at Tokyo Institute of Technology.
The combustion model was validated mainly by comparison with the experimental heat-release curves for 13 different cases of the parametric study reported by Kamimoto. A good match was demonstrated, though a weakness was revealed in the representation of the effect of air swirl on the jet. Otherwise, the combustion predictions were deemed to be sufficiently accurate to serve as an effective platform for the soot model study.
Soot formation was described using five different expressions, while six were used for the oxidation process. In general, it was found that the soot rate predictions were quantitatively poor, so that expression calibration will usually be necessary. Analysis of the predictions of exhaust smoke for different operating conditions revealed the importance of accurately describing the overall air-to-fuel ratio in the spray. The effect of load variation was poorly represented due to neglect of the transfer of combustion products between the model zones.
The comparative study of soot expressions identified a ranking of sensitivities of the formation expressions. Though oxidation is conceptually simpler, more distinct qualitative differences were observed in the behaviour of the expressions. For instance, under the high oxygen partial pressure conditions found in the diesel, the Nagle expression shows no sensitivity to composition at temperatures below 2000K. This gives a radically different behaviour to the simpler quasi-chemical expressions, such as those due to Lee and Hiroyasu. The effect on predicted exhaust levels is significant, since the final yield is dependent on the variation of the oxidation rate over the whole period of the combustion process. Thus, using certain expression combinations, a very high sensitivity to the soot model constants is shown.
Use of more detailed mechanisms, such as the model due to Tesner and Magnussen, was found to give no advantage. Similarly, the use of a more detailed underlying thermodynamic description makes it harder to achieve a good match of soot expressions. Generally, the best results can be obtained by use of simple quasi-chemical rate expressions.