Future emissions from maritime traffic will be limited by the IMO (International Maritime Organization) III regulation, which means further reductions of nitric oxide and soot emissions. Therefore, large gas engines become a genuine alternative to diesel engines operated with HFO (Heavy Fuel Oil).
A combustion process with high air/fuel-ratio enables low combustion temperatures and therefore low wall heat losses and nitric oxide raw emissions. On the other hand, the cycle-to-cycle variations increase, particularly for conventional ignition systems. Numerical methods can support the development process to define operating points with highest efficiencies and emissions that meet the regulation limits. However, these models must be able to describe the thermodynamics of the combustion process and the kinetically controlled knocking combustion and pollutant formation for different fuels.
Experimental investigations on the self-ignition process at different conditions and fuel mixtures were performed at the rapid compression machine in Aachen and it was found, that the mechanisms NATURAL GAS III from Galway describes the low and intermediate temperature chemistry very accurately. Based on reaction path and sensitivity analysis, the mechanism was reduced from 293 species to 60.
Additionally, a stochastic reactor model was set up in order to investigate the kinetically controlled processes of the knocking combustion and emission formation at transient conditions. Thus, the self-ignition timing in the unburned mixture was calculated for the knocking cycles in good correlation to the measurement.
Due to the fact that the knocking combustion and the emissions are closely related to the fast and slow burning cycles, an empirical relation for the cyclic variations was developed based on turbulent and chemical scales. Test bench data of a large gas engine with a prechamber spark plug was analysed with respect to the combustion statistics. Cycle-to-cycle parameters like the standard deviation of the duration of 0-10 % mass fraction burned were identified that determine the knock limit for a given operating point.
Afterwards, a 1D gas exchange model with a predictive combustion model was set up and adapted to the given engine. Therefore, the mean properties of the engine, like the fuel consumption and the power output, could be predicted. The empirical model for the cycle-to-cycle variations was implemented into the process simulation. The reduced reaction mechanism was transferred from the reactor model to the 1D gas exchange model. With this approach, engine knock and nitric oxide formation were predicted accurately for burning gases with different methane numbers.
The described model was used to optimize operating points with respect to highest efficiency and the constraint of a nitric oxide emission that meets different requirements. Engine parameters like inlet valve closing, spark timing, boost pressure, A/F-ratio and compression ratio created the parameter space. In order to search efficiently, a meta-heuristic optimisation algorithm was coupled with AVL BOOST (AVL List GmbH). A set of optimal operating parameters could be found for fuels with different methane numbers between 100 and 43 which enables low fuel consumption. It became obvious that high manifold pressures are suitable for high efficiencies. Nevertheless, an early IVC is the preferred measure to avoid knocking combustion for fuel mixtures with low methane numbers. The impact of different emission legislations was quantified, too. The stringent emission limits of the legislation TA Luft (Technical Instructions on Air Quality Control) decrease the efficiency noticeably. In this case, the positive effect of inert admixtures in the fuel could be confirmed.
The presented numerical approach showed great potential to identify the direction of future gas engine development. Therefore, it can play an important part for the future development of new gas engines with even higher efficiencies