Laminar flame speed modeling for a 1-D hydrogen combustion model
Examensarbete för masterexamen
Automotive engineering (MPAUT), MSc
CO2 emissions from internal combustion engines is a world wide problem by being one of the sources contributing to global warming. While the internal combustion engine has proven to be a reliable and versatile mobile power source, ranging widely in size and power output, the most commonly used fuels are of fossil origin introducing new CO2 to the atmosphere. Upcoming legislation’s will force new alternative fuels with reduced or zero CO2 emissions to the market. One of these alternative fuels is hydrogen which have the potential to be used in internal combustion engines. Volvo Penta have a lot of prior knowledge with diesel combustion but are taking a step into spark ignited engines with compressed natural gas (CNG) and hydrogen combustion to lower CO2 emissions. By the use of 1-D simulation the potential of hydrogen as a fuel in an internal combustion engine (ICE) can be evaluated. While 1-D simulation software like GTPower is commonly used when designing an ICE together with the traditional fossil fuels, gasoline and diesel, the potential to make simulations using hydrogen as the fuel is fairly unknown. For a predictive combustion model many of the combustion characteristics for the fuel is described by the laminar flame speed model. How fast the combustion occurs and therefore the energy release rate is partially determined by the laminar flame speed. Hydrogen has very high laminar flame speed at stoichiometric conditions compared to gasoline and diesel. In combination with high laminar flame speed the fuel can ignite at a wide range of equivalence ratios. Due to limits in material strength and a need for highly controlled combustion it is interesting to run hydrogen at lean conditions and by doing so limiting the laminar flame speed. By comparing the current laminar flame speed model used in GT-power to experimental data it was evident that the default laminar flame speed model did not give correct results near the lean limit for hydrogen. A chemical kinetics result based model was therefore created by using function fitting methods available in Matlab and then implemented into the predictive combustion model SITurb in GT-Power. The results when comparing the new kinetics-fit laminar flame speed model with the default GT laminar flame speed model showed similar results for lean combustion conditions until the very lean conditions occurred. For very lean conditions the kinetics-fit model showed more correlation to experimental results than the default GT-model did. While that was promising, the kinetics-fit model did not correlate well to the chemical kinetics results other than the points for which it was initially fitted for. In general the kinetics-fit model underestimated the laminar flame speed resulting in simulations showing inaccurate results.