Design and Simulation of Two Stroke Engines

Chapter 4 - Combustion in Two-Stroke Engines
theoretical approach, the use of empirically determined coefficients, particularly for factors
relating to turbulence, has increased greatly over the earlier proposal of a simple heat release
model to simulate the combustion process. It is somewhat questionable if the overall accuracy
of the calculation has been greatly improved, although the results presented by Reid [4.29-
4.31] are impressive. There is no doubt that valuable understanding is gained, in that the user
obtains data from the computer calculation on such important factors as exhaust gas emissions
and the flame duration. However, this type of calculation is probably more logical when
applied in three-dimensional form and allied to a more general CFD calculation for the gas
behavior throughout the cylinder leading up to the point of ignition. This is discussed briefly
in the next section.
4.4.4 Three-dimensional combustion model for spark-ignition engines
From the previous comments it is clearly necessary that reliance on empirically determined
factors for heat transfer and turbulence behavior, which refer to the combustion chamber
as a whole, will have to be exchanged for a more microscopic examination of the entire
system if calculation accuracy is to be enhanced. This is possible by the use of a combustion
model in conjunction with a computational fluid dynamics model of the gas flow behavior
within the chamber. Computational fluid dynamics, or CFD, was introduced in Chapter 3,
where it was shown to be a powerful tool to illuminate the understanding of scavenge flow
within the cylinder.
That the technology is moving toward providing the microscopic in-cylinder gas-dynamic
and thermodynamic information is seen in the paper by Ahmadi-Befrui etal. [4.21]. Fig. 4.10
is taken directly from that paper and it shows the in-cylinder velocities, but the calculation
holds all of the thermodynamic properties of the charge as well, at a point just before ignition.
This means that the prediction of heat transfer effects at each time step in the calculation will
take place at the individual calculation mesh level, rather than by empiricism for the chamber
as a whole, as was the case in the preceding sections. For example, should any one surface or
side of the combustion bowl be hotter than another, the calculation will predict the heat transfer
in this microscopic manner giving new values and directions for the motion of the cylinder
charge. This will affect the resulting combustion behavior.
This calculation can be extended to include the chemistry of the subsequent combustion
process. Examples of this have been published by Amsden et al. [4.20] and Fig. 4.11 is an
example of their theoretical predictions for a direct injection, stratified charge, spark-ignition
engine. Fig. 4.11 illustrates a section through the combustion bowl, the flat-topped piston and
cylinder head. Reading across from top to bottom, at 28° btdc, the figure shows the spray
droplets, gas particle velocity vectors, isotherms, turbulent kinetic energy contours, equivalence
ratio contours, and the octane mass fraction contours. The paper [4.20] goes on to show
the ensuing combustion of the charge. This form of combustion calculation is preferred over
any of those models previously discussed, as the combustion process is now being theoretically
examined at the correct level. However much computer time such calculations require,
they will become the normal design practice in the future, for computers are becoming ever
more powerful, ever more compact, faster and less costly with the passage of time.
323