The necessity for further reductions of in-cylinder pollutant
formation and the opportunity to minimize engine development and
testing times highlight the need of engine thermodynamic cycle
simulation tools that are able to accurately predict the effects of
fuel, design and operating variables on engine performance.
In order to set up reliable codes for indicated cycle simulation
in SI engines, an accurate prediction of heat release is required,
which, in turn, involves the evaluation of in-cylinder turbulence
generation and flame-turbulence interaction. This is generally
pursued by the application of a combustion fractal model coupled
with semi-empirical correlations of available geometrical and
thermodynamical mass-averaged quantities. However, the currently
available correlations generally show an unsatisfactory capability
to predict the effects of flame-turbulence interaction on burning
speed under the overall flame propagation interval.
Therefore, in the present paper, a new correlation that improves
the turbulent burning speed calculation is developed. It features
an original definition of the outer turbulence cutoff length scale,
based on the flame front area, and takes account of the increased
transfer across the flame front of both radical species and heat
for high in-cylinder densities. The correlation has been applied to
calculate the burning speeds in the cylinder of a naturally
aspirated bi-fuel engine for a wide range of engine speeds (N
2000-4600 rpm), loads (bmep 200-790 kPa), relative air-fuel
ratios (RAFR 0.80-1.30) and spark-advances (SA ranging from 8
deg retard to 2 deg advance with respect to MBT), under both
gasoline and CNG operations.
The computed burning speeds were then compared to those stemming
from the traditional correlation reported in the literature and to
the experimental flame propagation data. These latter were
extracted from the measured in-cylinder pressure traces by means of
a diagnostics technique previously developed by the authors. The
results indicate that the burning speeds calculated through the
authors' model are in better agreement with the experimental
outcomes than those derived from the traditional correlation widely
applied in the literature.