Changes in the engine operating conditions lead to changes in the output, which, by means of the turbocharger and the EGR, affect the engine input conditions。 Such changes require many engine cycles until the new equilibrium is reached。 The CFD-code utilized in this study simulates only the incylinder process from the inlet valve closure to the exhaust valve opening, assuming that the engine system is in equilibrium。 Therefore, in order to account for the EGR and the turbocharger feed-back, an additional simulation is required which can predict the correct engine input conditions for a given engine operating point。 This task is usually performed with a so-called zero-dimensional simulation code。 The zero-dimensional engine simulation code utilized in this study is a version of the OBEH code, developed at the Czech Technical University in Prague in the research group of J。 Macek。 This code has been designed primarily for turbocharged medium and large-bore DI diesel engines, equipped with an EGR system。 OBEH is interfaced with the CFD code KIVA-3 in such a way that a given engine operating condition, i。e, the injection timing, the EGR rate and the desired boost pressure, provides the appropriate equilibrium engine input data for KIVA-3 at inlet valve closure。 These input data include the temperature, pressure, and the cylinder composition due to EGR。 The OBEH models are outlined in the following。 A more detailed description can be found in [31]。 The highlights of the OBEH code include 

• a cumulative ignition delay model with a user defined function for induction time in a quasi-dimensional approach, 

• a simple empirical model for combustion with three cumulated Vibe functions based on Woschini’s ideas, 

• a heat transfer model with user defined heat transfer coefficients and an adaptive heat-resistance model, fitting surface temperatures of a piston, a cylinder and a head with valves, • a zero-dimensional exhaust manifold model calibrated by experiments, 

• a detailed turbine model using normalized maps for waste gate, variable geometry turbine (VGT) or by-pass boost control, two stage turbocharging, parallel and serial power gas turbine, mechanical or electrical super combined with a turbocharger, etc。 

• an EGR model for low and high pressure loops including venturis for pressure difference enhancement, 

• a simplified experimentally validated mechanical loss model which includes losses due to crank gears, piston, rings and bearings。

In addition, the code is equipped with many iterative control procedures used for the acceleration of the code calibration or optimization, e。g。, control of a constant peak pressure, VGT area control for constant boost pressure, air excess control etc。

MULTI-DIMENSIONAL ENGINE MODELS

The multi-dimensional engine computations for determining the pollutants and the power output have been performed with a modified version of the KIVA-3 code [27]。 This code is equipped with the RNG k−e turbulence model as implemented by Han and Reitz [32], the CAB atomization and drop breakup model [33, 34] and the LIT auto ignition model [35] of Tanner。 The heat release is modeled using the laminar-turbulent-laminar (LTL) characteristic time combustion (CTC) combustion model that is described in the following paragraphs, along with the emission models for soot and nitric oxide。 All other models used in the simulations are the standard KIVA-3 models。

The LTL-Characteristic Time Combustion Model

The LTL-CTC model is based on the CTC model of Abraham et al。 [36] as adapted to diesel combustion by Kong et al。 [37], but it employs only one global reaction to model the heat release。 As in the original CTC model, the LTL-CTC model uses a laminar reaction rate for the precombustion and a turbulence reaction rate for the spray combustion, but once the fuel injection is terminated, the LTL-CTC model gradually shifts back to the laminar reaction rate。 This last step is motivated by the fact that the turbulence in a diesel engine is dominated by the spray-induced flow during fuel injection, but it is considerably reduced in the later combustion phase, as is discussed in more detail in a study by Tanner and Reitz [38]。 The gradual shift from the mixing-controlled combustion back to the laminar combustion improves the reaction rate in the late combustion phase。 As a consequence, the notorious under-prediction of the heat release rate is improved, which results in a reduction of the unburned fuel at the end of the combustion。 Further details of the LTL characteristic time combustion model can be found in [39]。