Modeling Split Injections of ECN Spray A Using a Conditional Moment Closure Combustion Model with RANS and LES

This study investigates n-dodecane split injections of Spray A from the Engine Combustion Network (ECN) using two different turbulence treatments (RANS and LES) in conjunction with a Conditional Moment Closure combustion model (CMC). The two modeling approaches are first assessed in terms of vapor spray penetration evolutions of non-reacting split injections showing a clearly superior performance of the LES compared to RANS: while the former successfully reproduces the experimental results for both first and second injection events, the slipstream effect in the wake of the first injection jet is not accurately captured by RANS leading to an over-predicted spray tip penetration of the second pulse. In a second step, two reactive operating conditions with the same ambient density were investigated, namely one at a diesel-like condition (900K, 60bar) and one at a lower temperature (750K, 50bar). For the diesel-like condition, both modeling approaches captured first injection ignition delay, overall soot mass and soot location well. Due to an under prediction of spray tip expansion during combustion in combination with the slight over-prediction of the vapor tip penetration, RANS-CMC's reactive spray penetration length nonetheless agreed well with the experiment. However, RANS-CMC did not capture full combustion recession towards the injector location after the end of the first injection. LES-CMC showed accurate predictions of the second injection ignition delay and improvements with respect to combustion recession, and flame structure. For the low temperature condition, RANS-CMC accurately predicted ignition timing. Here LES-CMC slightly under-predicted ignition delay. Ignition location and flame structure are in reasonable agreement with the experiment for both numerical setups. In accordance to the experiment it was seen that for both RANS-and LES-CMC the first injection does not auto-ignite and the second injection was necessary to create the appropriate local conditions for ignition to occur. Furthermore, the simulations did not predict any soot at the lower temperature, in agreement with the experimental observations. The findings suggest that LES may be necessary in the case of split injections towards accurate two-phase flow field predictions and - even more so - to capture full combustion recession for diesel-like conditions. The proposed RANS-and LES-CMC models show considerable promise for the challenging dynamics of auto-igniting fuel sprays with multiple injections.