Search Results

The basic idea behind large-eddy simulation (LES) is to accurately resolve the large energy-containing scales and to use subgrid-scale (SGS) models for the smaller scales. The accuracy of LES can be significantly impacted by the numerical discretization schemes and the choice of the SGS model. This work investigates the accuracy of low-order LES codes in the simulation of a turbulent round jet which is representative of fuel jets in engines. The turbulent jet studied is isothermal with a Reynolds number of 6800. It is simulated using Converge, which is second-order accurate in space and first-order in time, and FLEDS, developed at Purdue University, which is sixth-order accurate in space and fourth-order in time. The high-order code requires the resolution of acoustic time-scales and hence is approximately 10 times more expensive than the low-order code.

In this work, computations of reacting diesel jets, including soot and NO, are carried out for a wide range of conditions by employing a RANS model in which an unsteady flamelet progress variable (UFPV) sub-model is employed to represent turbulence/chemistry interactions. Soot kinetics is represented using a chemical mechanism that models the growth of soot precursors starting from a single aromatic ring by hydrogen abstraction and carbon (acetylene) addition and NO is modeled using the kinetics from a sub-mechanism of GRI-Mech 3.0. Tracer particles are used to track the residence time of the injected mass in the jet. For the soot and NO computations, this residence time is used to track the progression of the soot and NO reactions in time. The conditions selected reflect changes in injection pressure, chamber temperature, oxygen concentration, and density, and orifice diameter.

Accurate modeling of the transient structure of reacting diesel jets is important as transient features like autoignition, flame propagation, and flame stabilization have been shown to correlate with combustion efficiency and pollutant formation. In this work, results from Reynolds-averaged Navier-Stokes (RANS) simulations of flame lift-off in diesel jets are examined to provide insight into the lift-off physics. The large eddy simulation (LES) technique is also used to computationally model a lifted jet flame at conditions representative of those encountered in diesel engines. An unsteady flamelet progress variable (UFPV) model is used as the turbulent combustion model in both RANS simulations and LES. In the model, a look-up table of reaction source terms is generated as a function of mixture fraction Z, stoichiometric scalar dissipation rate Xst, and progress variable Cst by solving the unsteady flamelet equations.

The following computational study examines the structure of sonic hydrogen jets using inlet conditions similar to those encountered in direct-injection hydrogen engines. Cases utilizing the same mass and momentum flux while varying exit-to-chamber pressure ratios have been investigated in a constant-volume computational domain. Furthermore, subsonic versus sonic structures have been compared using both hydrogen and ethylene fuel jets. Finally, the accuracy of scaling arguments to characterize an underexpanded jet by a subsonic “equivalent jet” has been assessed. It is shown that far downstream of the expansion region, the overall jet structure conforms to expectations for self-similarity in the far-field of subsonic jets. In the near-field, variations in fuel inlet-to-chamber pressure ratios are shown to influence the mixing properties of sonic hydrogen jets. In general, higher pressure ratios result in longer shock barrel length, though numerical resolution requirements increase.

This paper provides a comparison of wall heat fluxes and quenching distances as one-dimensional hydrogen and heptane flames impinge head-on onto a wall. It is shown that the quenching distances for stoichiometric H2/air and C7H16/air flames under the specified conditions of this study are about the same, but the wall heat flux for the H2/air flames is approximately a factor of two greater. For lean H2/air mixtures, the quenching distance increases substantially and the wall heat flux decreases. To understand more clearly the interplay of flame speed, temperature, thermal diffusivity, and surface kinetics on the results, studies of H2/O2 flames are also carried out.

A multidimensional model is employed to model ignition and heat release rates in a Diesel engine. An interactive flamelet model is employed to model combustion. Nheptane is used as a representative fuel for Diesel fuel in the computations. Comparisons of computed and measured results are presented for a range of engine operating conditions: speed 1200 rpm, start of injection 12.5 degrees before top dead center to 9.5 degrees after top dead center and intake air temperature of 340-360 K. The primary objective of this work is to assess the ability of the model to reproduce ignition timings. The flamelet model uses detailed chemical kinetics and it is shown that it can reproduce the qualitative trends of changes in ignition delay and heat release rates with respect to changes in operating conditions of the engine. The capability to reproduce the measured changes in ignition delay is important because changes in injection timing lead to changes in ignition timing.

In Diesel engines, the vapor phase of the fuel jet is known to impinge on the walls. This impingement is likely to have an effect on mixing characteristics, the structure of the diffusion flame and on pollutant formation and oxidation. These effects have not been studied in detail in the literature. In this work, the structure of a laminar wall jet that is generated from the impingement of a free laminar jet on a wall is discussed. We study the laminar jet with the belief that the local structure of the reaction zone in the turbulent reacting jet is that of a laminar flame. Results from non-reacting and reacting jets will be presented. In the case of the non-reacting jets, the focus of the inquiry is on assessing the accuracy of the computed results by comparing them with analytical results. Velocity profiles in the wall jet, growth rates of the half-width of the jet and penetration rates are presented.

In this paper, results from an experimental and computational study relating NO and soot emissions in a Diesel engine to heat release rate characteristics are reported. The experiments were carried out in a Cummins N-14 single-cylinder Diesel engine. The computations were carried out for the same engine. It is shown that, more than any significant feature of the heat release rate itself, the NO appears to be related to the temperature of the reactants with higher temperatures resulting in higher NO emissions. Relationships of NO to the heat release rates are secondary to this primary dependence. In general, the soot-NO trade-off relationships appear to hold. However, for the range of conditions studied, soot and NO are found to simultaneously decrease with decreasing air temperatures. It is also found that at the most retarded timings, NO and soot simultaneously decrease but with a severe penalty in fuel consumption.

In this work, a multidimensional model that the authors have previously developed for computing radiant heat loss in an internal combustion engine is applied to study radiant heat loss characteristics in a Diesel engine. The model is applied to study the effects of load and speed on radiant heat transfer in the engine. It is shown that as load is increased the radiant heat loss increases and the fraction of radiant to total heat loss increases from about 12% at an overall equivalence ratio of 0.3 to 16% at an overall equivalence ratio of 0.5. As speed is increased, the radiant and total heat loss again increase but the ratio of radiant to total heat loss remains about the same for the cases considered. It is shown that there is a strong correlation between the radiant heat loss characteristics and soot concentration and temperature in the chamber.

The spray submodel is an important component in multidimensional models for Diesel engines. The satisfactory representation of the spray is dependent on adequate representation of turbulence in the jet which, in part, determines its spread and penetration. In this work, the RNG k-ϵ model is evaluated relative to the standard k-ϵ model for computing turbulent jets. Computations are made for both gas jets and sprays. The gas jet is computed with an adequately high degree of numerical spatial resolution of the order of the orifice diameter. In the case of the spray, achieving such a high resolution would be challenging. Since the spray has similarities to the gas jet, and the gas jet may be computed with such high resolution and adequate accuracy, firm conclusions may be drawn for it and they may be applicable to sprays. It is concluded that the RNG k - ϵ model, in general, results in predictions of greater mixing in the jets relative to the standard model.

It is generally agreed that adequate resolution is required to reproduce the structure of spray and gas jets in numerical computations. It has not been clarified what this resolution should be although it would appear reasonable to assume that it should be such that the physical scales of the problem are resolved. In the case of a jet, this implies that near the orifice, the jet diameter has to be resolved since this is the appropriate length scale. It is shown in this work that if such a resolution is not used in computing transient jets, the structure of the jet is not reproduced with adequate accuracy. In fact, unexpected, erroneous and misleading dependence on ambient turbulence length and time scales will be predicted when the initial ambient turbulence diffusivity is small relative to the jet diffusivity. When the ambient turbulence diffusivity is of the same order as the jet diffusivity or greater, entrainment rates are significantly underpredicted.

The first comparisons of measured and computed mean velocities, and of measured fluctuation intensities and computed turbulence intensities in a motored rotary engine are presented. The computations were performed with a recently developed three–dimensional model. The measurements were made at the Sandia National Laboratories by Dimpelfeld and Witze at several locations along the rotor housing and at two engine speeds. The measured and computed mean velocities agree to within 15% whereas the computed turbulence intensities correctly are lower than the measured fluctuation intensities by about 30% to 50%, as anticipated. Under motored conditions, the turbulence intensity tends to be rather homogeneous and of similar magnitude somewhat before and after top dead center but significantly inhomogeneous and of greater magnitude around top dead center. The comparisons suggest that predictions of mean gas velocity and of turbulence intensity can be made with the available model.

Propagation of turbulent flames in spark-ignition engines is considered from the viewpoint of the different possible regimes of premixed turbulent combustion. Nondimensional parameters defining known combustion regimes are reviewed, and numerical values of these parameters are estimated for both research and production engines. The reaction-sheet regime is inferred to apply at least for some operating conditions, and therefore literature on turbulent flame propagation in the reaction-sheet regime is reviewed. Implications of these results on interpretations of existing experimental observations of combustion in engine cylinders and on modeling of turbulent flame propagation in engines are discussed.