Emission Models

In diesel engines, nitric oxide formation is dominated by the high-temperatures。 This high-temperature NOx formation is modeled using the extended Zeldovich mechanism is described in detail in [25]。 The forward and the backward reaction rates of the Zeldovich mechanism as a function of the temperature, T, are computed from the Arrhenius relation 

where the constants K and b along with the activation temperature To are listed in Table 4。 The H, O and OH radicals involved in the Zeldovich mechanism are obtained from the set of equilibrium reactions listed in Table 5。 In this table the constants{A,B,C,D,E}are the ones in the expression for the equilibrium concentration Ke, which is given by 

where Tr = T/1000。 For additional details see [40]。 The net soot density is modeled according to Hiroyasu and Kadota [41] by 

where ρs is the net soot density, ρsf is the density of soot formed and ρso is the density of soot oxidized。

The soot formation is modeled according to (cf。 [41])

where ρv is the fuel vapor density, p is the pressure in bar, T is the temperature in Kelvin, Ef=52300 J/mol is the activation energy, and R=8。3143 J/(mol K) is the universal gas constant。 Cf is a tuning constant and the value used in this study is Cf = 4。7。 The oxidation model used is the one by Nagle and Strickland-Constable (NSC) [42], as presented by Chan et al。 [10]。 The overall oxidation rate of the chemical reactions represented by this model is given by

where Co=1。215 is a tuning constant, mc=12 g/mol is the molecular weight of carbon, σ=2 g/cm3 is the density of a soot particle, and d = 3×10−6 cm is the average diameter of a soot particle。 Rt represents the overall reaction rate of this system。 Further details can be found in [39]。

COMPUTATIONAL DETAILS

The optimization algorithm has been programmed in C。 This code also served as the fully automated control interface for the optimization runs。 The zero-dimensional engine simulations were used to predict the gas composition at the closure of the inlet valves。 This data is then used in the threedimensional engine simulations, that have been performed with a KIVA-3-based code using the models described in the previous section, to determine the emissions and SFC。 The engine used in the simulations is the four-stroke Sulzer S20 DI diesel engine with a central injector equipped with 12 nozzle orifices。 The experimental data used in the tuning case has been obtained from a stationary nine-cylinder production engine [43]。 The engine specifications are listed in Table 6。 The cylinder flow is assumed to be periodic with respect to the number of nozzle orifices and hence only one sector of the engine cylinder corresponding to one injection orifice is simulated。 The mesh used for the engine simulation had 23×7×14 cells in radial, azimuthal and vertical direction, respectively at TDC。 This corresponds to a computational domain of 30deg。 As discussed at the end of the Model Tuning subsection, this mesh gave adequate mesh independent results。 All the computations have been performed from the closure of the inlet valves at 144 CA before TDC to the opening of the exhaust valves at 129 CA after TDC。 A common rail injection system was used which was simulated with constant injection。

摘要：本研究的目的是找到最优工况的燃油消耗和排放的高效计算 CFD的工具。优化算法是基于梯度下降的方法。自适应最小化代价函数使用有效回溯沿线搜索。自适应的代价函数是基于罚函数法，每条线搜索后罚函数都会增加。因此参数空间的规范发生在高维空间中的优化单位立方体。这种优化工具的应用被应用于Sulzer S20，这是一个中央喷射非管路柴油发动机。优化参数包括双脉冲的开始时间，每个脉冲的持续时间，停留时间，废气再循环率和升压压力。零维发动机代码是在关闭的入口阀门用来模拟预测排气和进气冲程。这些数据将作为初始值进行了三维数值模拟，用来计算的排放量和燃油消耗。仿真是进行测试不同功耗的不同油耗。最好的情况表明，一氧化氮和颗粒物可降低83%以上，至少接近24%，分别低于EPA的任务标准，同时保持一个合理的规范燃料消耗值。此外，采取起点算法最佳的路径进行了研究，了解各参数优化过程的影响。来,自.优;尔:论[文|网www.youerw.com +QQ752018766-