The heat release of the low temperature reactions (LTR or cool-flame) under Homogeneous Charge Compression Ignition (HCCI) combustion has been quantified for five candidate fuels in an optically accessible Rapid Compression Expansion Machine (RCEM). Two technical fuels (Naphthas) and three primary reference fuels (PRF), (n-heptane, PRF25 and PRF50) were examined. The Cetane Numbers (CN) of the fuels range from 35 to 56. Variation of the operating parameters has been performed, in regard to initial charge temperature of 383, 408, and 433K, exhaust gas recirculation (EGR) rate of 0%, 25%, and 50%, and equivalence ratio of 0.29, 0.38, 0.4, 0.53, 0.57, and 0.8. Pressure indication measurements, OH-chemiluminescence imaging, and passive spectroscopy were simultaneously implemented.
In our previous work, an empirical, three-stage, Arrhenius-type ignition delay model, parameterized on shock tube data, was found to be applicable also in a transient, engine-relevant environment. The pressure rise due to cool-flame heat release, which is crucial for the induction of main ignition, was included in the experimental pressure traces that have been used. To fully predict the ignition delay in HCCI-engine applications however, the cool-flame heat release characteristics need to be known in advance.
In this work, the cool-flame heat release characteristics have been investigated with regard to operating parameters. A simplified, cool-flame heat release model is proposed, that is mathematically independent from the three-stage ignition delay model. It provides the cool-flame heat release profile that is used to reconstruct a pressure/temperature trace including the effect of the cool-flame. The reconstructed trace is the input to the three-stage model, and thus both cool and hot-flame ignition delays can be predicted. The overall performance of the combined three-stage/cool-flame heat release model was assessed. Very good agreement was observed between the experimental ignition delay and the combined cool-flame/three-stage ignition model computations.