A study is presented on the modeling of fuel spray combustion in diesel engines. The objective is to model igniting diesel sprays with the detailed chemistry tabulation method FGM (Flamelet Generated Manifold). The emphasis is on the accurate prediction of auto-ignition as well as the steady combustion phase using one consistent approach. Introduction with strain rates varying from a low value (close to Due to ever increasing demands from emis- equilibrium) to the quenching value is computed. sion legislation (NOx and soot), fuel economy (CO2 An illustrative example of the ”accessible” space in ) and fuel flexibility (bio-fuels) diesel engines be- Z-P V is shown in Figure 1, see the gray area become more and more complex. Therefore, conven- tween the solution for the lowest strain rate and the tional engine design approaches that rely on proto- solution at which the strain rate reached its maxitype development become too time-consuming and mum before extinction. expensive. The development of predictive and efficient computational tools would represent a sig- Flamelet database generation nificant step forward in the ability to rapidly design high efficiency, low emission engines [1]. Modern diesel engine technology unequivocally applies liquid fuel injection with high pressure, that forms a non-homogeneous mixture leading to relateivxetrleymheiglyh dleifvfieclusltodfuseootot. tMheodceolminpglethxisphperoncoemsesnias []PV occurring during fuel injection and combustion. A direct approach (DNS) is unviable and advanced, extinguishing flamelet so accurate and fast sub-models are needed to apply it in engine design. igniting flamelet Recently, efforts to accurately and efficiently igniting homogeneous reactors model diesel spray formation resulted in a suitable model to compute the mixture formation [2]. And 0 0.1 0.2 0.3 0.4 Z0[.5−] 0.6 0.7 0.8 0.9 1 a first application of the FGM method to a diesel Lowest strainrate spray proved that experimentally observable phe- SHtigehaedsytsnoolunt−ioqnusenrecghiionng strainrate nomena like auto-ignition and flame lift-off can be THiommeodgeepneenoduesntrleyaecxttoinrsgubiesfhoirnegigonriitgionniting flamelet successfully predicted with this method [3]. The objective of this study is to compare the spray igni- Figure 1: Ways to extend a stationary database tion behavior for different manifold types.
The FGM approach combines the flamelet con- Diesel combustion is characterized by
autocept with a manifold method by using mixture frac- ignition, so to cover this aspect the table should
tion Z and progress variable P V to parameterize also contain information in the area beneath the
the combustion process [
FGMs can be generated in many ways. For sta- flamelets from a mixed, but non-reacting initial
tionary flames, there is a classical way with steady state. The ignition behavior is followed in time until
flamelets only, where a sequence of steady flames a steady flame is reached. A third possibility is to
reproduce ignition of mixtures covering the entire
Z-space with homogeneous reactor auto-ignition
calculations [
Due to the unsteady nature of a diesel injection event, ignition modeling is at least as important as combustion modeling. Following the FGM approach, besides combustion, ignition should be covered inherently. But the result depends on the way the FGM is generated. The extinguishing flamelet approach is applied and does not lead to ignition of the spray. Instead, only local temperatures slightly above the initial ambient temperature are found, and the source of the reaction progress variable is not big enough to end in total ignition within a few milliseconds. However, a FGM constructed with an igniting flamelet or homogeneous reactor database does result in auto-ignition of the Figure 2: Igniting flamelet manifold. P V source as funcwhole spray in short time. Therefore, in this paper tion of Z and P V . only the results of the igniting flamelet and homogeneous reactor approaches are presented.
In this study non-premixed flamelets for a
counterflow setup are solved with CHEM1D [
Here the Z definition of Bilger [9] is adopted
and P V is chosen as a combination of CO2, CO
and CH2O mass fractions. The created laminar
manifolds parameterized with the mixture fraction Figure 3: Homogeneous reactors manifold. P V source
Z and reaction progress variable P V are depicted as function of Z and P V .
in Figure 2 for the flamelets case (ignition at strain
rate 500) and in Figure 3 for the homogeneous
reactors case. In these two figures the contours of tors case is at a more fuel rich region (Z > Zstoich)
the P V source is plotted. Some major dissimilari- than in the flamelet case (Z Zstoich = 0:064).
ties can be observed when these figures are com- These are also visualized in Figure 4 by means
pared. Firstly, the position of the maxima of the of temperature contours at the moment that an
igsources is different, and secondly the values of the nition spot increases with 50 K above the
ambimaxima differs with a factor of approximately 30. ent temperature, prior to an exponential increase
The igniting flamelet database shows high source to temperatures above 2000 K. So far, when
comterms in the region 0:05 < Z < 0:15, whereas pared with literature [
Eventually, when the laminar manifolds are integrated with presumed -PDF functions, the dissimilar manifolds also result in different auto-ignition position and time of a 3D turbulent spray. Due to the higher P V sources, the homogeneous reactors Figure 4: Temperature contours. At start of injection the lead much faster to ignition than the flamelet case. time is 0 ms and the ambient temperature is 800 K. And because of the position of the high source terms, the ignition spot in the homogeneous reacOutlook heptane. Combustion and Flame, 128:38–59,
In the quest for a generic approach to model 2002. the important chemistry related characteristics in [9] R.W. Bilger, S.H. Starner, and R.J. Kee. On rediesel sprays, the observed dissimilarities in auto- duced mechanisms for methane-air combusignition prediction between the FGM tabulation tion in nonpremixed flames. Combustion and strategies are to be investigated. This is going to Flame, 80:135–149, 1990. be done with 0D and 1D laminar simulations to ex- [10] Dean Verhoeven, Jean-Luc Vanhemelryck, clude the possible influences of turbulence mod- and Thierry Baritaud. Macroscopic and igeling and/or complicated interactions within the nition characteristics of high-pressure sprays used Fluent framework. Especially the effects of of single-component fuels. SAE paper, (SAE progress variable choice and the choice of strain 981069), February 1998. rate at which the igniting flamelet is calculated are points of interest.
This project is funded by the Dutch technology foundation STW, and involves the following industrial partners: Shell Global Solutions and DAF Trucks. Their contribution is greatly acknowledged.