Advanced turbulence and combustion modeling for the study of a swirl-assisted natural gas spark-ignition heavy-duty engine

Increasing demands on higher performance and lower fuel consumption and emissions have led the path for internal combustion engine development; this race is nowadays directly related to CO2 emissions reduction. To drive engine development and reduce the time-to-market, the employment of numerical analysis is mandatory. This requires a continuous improvement of the simulation models toward real predictive analyses able to reduce the experimental R&D efforts. In this framework, 1D numerical codes are fundamental tools for system design, energy management optimization, and calibration. The present work is focused on the improvement of the phenomenological turbulence model, originally conceived to describe turbulence evolution in tumble-promoting engines. The turbulence model is developed with reference to a SI heavy-duty CNG engine derived from a diesel engine. In this architecture, due to the flat cylinder head, turbulence is also generated by swirl and squish flow motions, in addition to tumble motion. The presented turbulence model is validated against 3D CFD results, demonstrating to properly predict turbulence and swirl/tumble evolution under two different operating conditions, without the need for any case-dependent tuning. The developed turbulence model is coupled to a phenomenological combustion model based on the fractal geometry theory applied to the flame front surface, where the turbulence is assumed to support flame propagation through an enhancement of the flame front area with respect to the laminar counterpart. The combustion model is validated against an extensive experimental dataset, composed of 25 operating points at different engine rotational speeds and loads. The numerical/experimental comparisons of global performance parameters are satisfactory, leading to maximum errors around ±2% for the BSFC, ±2° for the main combustion events, and ±1 bar for the in-cylinder peak pressure. Burn rate profiles are very well captured by the combustion model at changing operating conditions, not requiring any case-dependent tuning. The presented results demonstrate that the turbulence/combustion models could constitute a reliable virtual test facility, contributing to supporting and driving experimental activities.