Research in Engineering and Aviation
Combustion-response mapping procedure for internal-combustion engine emissions
Authors: Korakianitis, T., Imran, S., Chung, N., Ali, H., Emberson, D.R., Crookes, R.J.
Applied Energy, Vol. 156, pp. 149-158. October 2015. DOI: 10.1016/j.apenergy.2015.06.039
This paper describes a new method to predict emissions in internal combustion (IC) engines. The method couples a multi-dimensional engine modeling program with pre-integrated non-equilibrium chemical kinetics reaction results. Prior to engine simulation, detailed chemical kinetics reactions of air/fuel mixture at different temperatures, pressures, and compositions, are calculated using SENKIN, a subprogram in the CHEMKIN-II computer package. The reaction results are decoupled from their chemical eigenvalue (order of about 10-1010-10 s), then integrated and saved in physical time scale (order of about 10-510-5 s) in a database file. In the database reaction results of different initial conditions (temperature, pressure, and species composition) are stored in different zones; the zones are indexed using their respective reaction conditions. Fluid dynamics and thermal dynamics of the movement of piston and valves, and spray droplets interaction are simulated by KIVA-3 V. Instead of calculating directly the non-equilibrium chemical reactions of the air/fuel mixture, reaction results are obtained from the database file via an interpolating subroutine, which returns temperature, heat release, and species concentrations after reaction to the main program. The approach avoids direct time consuming calculation of detailed chemical reactions as well as the errors introduced by coupling the physical and chemical processes. Emissions are predicted accurately since reaction of air/fuel mixture is calculated using the detailed chemical kinetics mechanism. The approach is applied to model a Caterpillar 3401 direct injection compression ignition (CI) diesel engine. In addition we carried out experimental tests on a Toledo 1500 SI gasoline engine, and those results are compared with the proposed computational approach. In all cases the predicted results agree well with the experimental data.