Design and Simulation of Two Stroke Engines

1.4
= 1 3
OC
LU
DC
Z)
CO
CO
LU
DC
Q.
1.2 -
1.1
AIR
Chapter 2 • Gas Flow through Two-Stroke Engines
CALCULATED
1.0
TIME, seconds
0.9
0.00 0.01 0.02 0.03 0.04 0.05 0.06
Fig. 2.57 Measured and calculated pressures at station 3.
to station 1, is 0.03 second. Very approximately and ignoring interference due to superposition
as in Eqs. 2.2.9 and 2.2.10, in a pipe filled with air only, that return time would have been:
return time, t
t = 2(5.913-0.317)
421.7
= 0.0265 second
As the peak of the exhaust pulse passed station 1 at 0.004 second originally, it would have
returned there at 0.0265 + 0.004, or 0.0305 second, and not at the observed value of 0.035
second. The phase error emanating from a simulation process detailing a completely air-filled
pipe would have been highly visible in Fig. 2.55.
In fact, in Figs. 2.55 to 2.57 is a third pressure trace where the computation has been
conducted with the pipe filled only with air. The traces are marked as "air" or "air only." The
phase error deduced very simply above is seen to be remarkably accurate, for in Fig. 2.55 the
"air only" returning reflection does arrive at station 1 at 0.031 second. On the other graphs the
"air only" computation reveals the serious error that can occur in UGD simulation when the
correct properties of the actual gases involved are not included. Needless to add, in Fig. 2.55
there is no sign of the pressure wave "bump" at 0.025 second for there is no CO2 in that
computation!
From the measured and computed pressure traces it can be seen that the waves have
steep-fronted and that the computation follows that procedure very accurately. Some phase
error is seen by 0.047 second at station 3. This is after some 15.5 meters of pressure wave
propagation.
It can be observed that the GPB finite system modeling gives a very accurate representation
of the measured events for the reflection of compression pressure waves at gas property
variations within ducting.
2.20 Computation time
One of the important issues for any computer code is the speed of its operation. Kirkpatrick
et al. [2.41] conclude that the GPB finite system simulation method and the Lax-Wendroff
(+Flux Corrected Transport) have equality of computational speed and both are several times
191