A crank angle resolved CIDI engine combustion model with arbitrary fuel injection for control purpose

With the introduction of new generation common rail system Diesel engines capable of multiple injections per stroke, and in the context of ever more stringent pollutant emission standards, the optimization and calibration of modern compression ignition direct injection (CIDI) engines is more and more complex. On one hand, the additional degrees of freedom provide additional opportunities to optimize the engines. On the other hand, the additional flexibility does not permit the exhaustive engine mapping approach used in the past any more, and necessitate the advent of new modeling tools for rapid optimization and calibration. Those tools must be sufficiently accurate to capture all the relevant physical phenomena, yet simple enough to be computationally cheap and permit the exhaustive exploration of a large multi-dimensional design and control space. In particular, one of the missing such model today is a simple and efficient tool for CIDI combustion simulation with arbitrary fuel injection profile. Thus in this study, a crank-angle resolved CIDI engine combustion model was developed and validated. This model uses a single-zone approach and is limited to the closed-valve part of cycle of a single cylinder for computational efficiency. All these sub-models were suitably parameterized in terms of externally controllable variables. To perform these parameterizations and sub-model validations, extensive experiments were conducted using a fuel injection rig and a multi-cylinder engine on a dynamometer. With this combustion model, the crank-angle resolved history of the in-cylinder parameters was calculated. Based on these results, the NOx emissions were predicted using the extended Zeldovich mechanism, and applying the concept of local equivalence ratio to calculate the temperature of each burned gas element.; To validate the overall combustion and NOx estimation models, a series of engine tests were performed over a range of operating conditions. The calibrated models allow to accurately predict in-cylinder pressure, torque and NOx emissions and also allow to perform a virtual dynamometer mapping, hence demonstrating the proposed methodology. Furthermore, the results from these virtual engine mappings can be used to calibrate the black-box models of the combustion process which are typically used in control-oriented, dynamic Mean Value Models.