Low-temperature ignition behavior of iso-octane

Auto-ignition properties of iso-octane mixtures were systematically investigated at conditions relevant to practical combustion devices using the University of Michigan Rapid Compression Facility and the Tsinghua University Rapid Compression Machine. Pressure time history measurements and high-speed imaging of the ignition process were used in both facilities to determine auto-ignition delay times and directly observe physical ignition behaviors. Test mixtures used fuel-to-O2 equivalence ratios of φ = 0.25 and 1.0, and were air-dilute, i.e. molar O2 to diluent gas (N2, Ar) ratio of 1:3.76. The pressures and temperatures after compression ranged from 3 to 30 atm and 740-1125 K respectively. The comprehensive results of the present work combined with those from previous shocktube studies clearly illustrate the existence of both inhomogeneous and homogeneous auto-ignition behaviors at these conditions. Analysis of patterns in the ignition behaviors revealed a dependence on initial unburned temperature and pressure, as well as equivalence ratio, with distinct regions of thermodynamic state in which the behavior is consistent and repeatable. The strong ignition limits were identified for both φ using the experimental results and compared to predicted locations made using the Sankaran Criterion (the ratio of the laminar flame speed to the thermal gradient driven spontaneous propagation speed). Predictions made using an assumed thermal gradient of 5-10 K/mm were in excellent agreement with measurements at all conditions, clearly indicating that use of this criterion is an effective method for a priori prediction of auto-ignition behaviors for iso-octane. This validation of the Sankaran Criterion for iso-octane, an important reference hydrocarbon fuel, importantly broadens the use of this tool and is an indication that ignition processes in hydrocarbon and high hydrogen content fuels are fundamentally similar. Additionally, a comparison of the measured auto-ignition delay times to predictions made using zero-dimensional homogeneous reactor modeling revealed that for experiments with inhomogeneous ignition behaviors, agreement was dependent on φ and the auto-ignition delay time. The presence of inhomogeneous ignition behavior did not significantly affect the accuracy of auto-ignition delay time predictions for mixtures with φ = 0.25; whereas, for mixtures with φ = 1.0 the presence of inhomogeneous ignition behavior significantly reduced the accuracy of predictions if the auto-ignition delay time was greater than ∼1 ms. These results indicate that while lowering φ may not eliminate inhomogeneous ignition behaviors, the subsequent effect of these behaviors on the predictive accuracy of typical zero-dimensional ignition modeling can be reduced or eliminated.