This research was conducted in two phases: an experimental investigation of diesel engine combustion and computational model development and validation. In the first phase, experiments were conducted to compare three alternative low-temperature combustion (LTC) strategies to two conventional high-temperature combustion (HTC) conditions in a single-cylinder optical engine. These combustion strategies were investigated using a number of optical diagnostics including two-color soot thermometry, line-of-sight soot luminosity imaging, line-of-sight chemiluminescence imaging, planar laser induced fluorescence (PLIF) of OH and fuel, and planar laser induced incandescence (PLII) of soot. Conceptual models that highlight important characteristics of each operating condition were proposed and discussed in detail. The conceptual models for the LTC conditions were found to be substantially different from those at HTC conditions. For the HTC conditions, soot is formed from the upstream to mid-stream regions of the fuel jet that is surrounded by a relatively thin OH distribution at the periphery of the jet. However, for the LTC conditions the soot was found to exist primarily in the head vortices of the jet, and OH was broadly distributed in the downstream regions of the jet.
In the second phase, four different combustion models, a characteristic time combustion (KIVA-CTC) model, a representative interactive flamelet (KIVA-RIF) model, direct integration using detailed chemistry (KIVA-CHEMKIN) model, and a hybrid auto-ignition/flame-propagation (KIVA-CHEMKIN-G) model were validated against data obtained from the experimental investigation. The models were integrated into the same version of the KIVA-3V computer code. Comparison of model predictions with the experimental results shows that all the models perform reasonably well in predicting the cylinder pressure and heat release rate. The KIVA-CHEMKIN and KIVA-CHEMKIN-G models better predicted the in-cylinder details for all the five operating conditions for the diesel engine.
The models were also validated with the Sandia flame liftoff length experiments and with diesel/natural-gas dual-fuel engine experiments. The KIVA-RIF model does not predict a lifted diesel flame. The KIVA-CHEMKIN and KIVA-CHEMKIN-G models give good qualitative predictions of the trends in liftoff. The newly proposed KIVA-CHEMKIN-G model performed the best in predicting trends in cylinder pressure and NOx emissions for the dual-fuel engine.