Based on experimental characterization of the scavenging behavior of a cross-scavenged, piston-aspirated, two-stroke, natural gas engine in phase III of the current project, a computationally inexpensive simple scavenging model was improved in this phase. Experimental results using fast nondispersive infrared (NDIR) CO2 measurements from the cylinder and the exhaust, as well as experiments using unburned fuel pre-mixed in the scavenging chamber as a tracer for short-circuiting during scavenging, were used in this phase to validate the improved model. The model represents the fundamental phenomenological...
Based on experimental characterization of the scavenging behavior of a cross-scavenged, piston-aspirated, two-stroke, natural gas engine in phase III of the current project, a computationally inexpensive simple scavenging model was improved in this phase. Experimental results using fast nondispersive infrared (NDIR) CO2 measurements from the cylinder and the exhaust, as well as experiments using unburned fuel pre-mixed in the scavenging chamber as a tracer for short-circuiting during scavenging, were used in this phase to validate the improved model. The model represents the fundamental phenomenological characteristics revealed by those experiments. The experiments and literature show that scavenging takes place by the following phenomena: blowdown, displacement of residuals by incoming air, mixing of residuals and air, and short-circuiting of fresh air. To reflect this, the improved hybrid model features the following: isentropic blowdown, non-isothermal perfect displacement, non-isothermal perfect mixing, and a concurrent direct short-circuiting of air (unmixed with residuals).
The validated improved hybrid model rectified the primary shortcoming of the phase III model. By adding the discrete short-circuiting zone, trapped mass could be modeled at both medium and high crankshaft speeds, whereas the phase III model could not capture the full scope of scavenging inefficiencies at medium speed using its perfect mixing stage alone. Furthermore, using the hybrid model to predict NOx using an exponential NOx-TER curve fit revealed that the improved phase IV hybrid model predicts NOx approximately as well as the experimentally-calculated TER from the phase III experiments.
Additionally, GT-Power, a 1D fluid dynamics and engine simulation software, was used to identify whether hybrid model tuning could be aided by relatively inexpensive 1D simulation rather than CFD or fast NDIR experiments. Using three-pressure analysis (with in-cylinder, exhaust, and scavenging chamber pressures as boundary conditions) and scavenging profiles derived from the hybrid model itself, GT-Power was shown to be a plausible tool for scavenging model tuning.