Search Results

The conditional moment closure (CMC) model has successfully been employed for simulation of turbulent nonpremixed combustion in diverse engineering problems. The solution procedure was developed for Lagrangian CMC with a single or multiple fuel or flame groups for a diesel engine in our previous work. Multiple flame groups may consider different histories and residence times through the sequence of evaporation, ignition and mixing controlled combustion. The Lagrangian CMC model has an advantage of less computational burden than Eulerian CMC with fixed spatially resolved grids. However, it cannot allow interaction among different flame groups through premixed combustion, while such interaction occurs through spatial transport terms in Eulerian CMC. In this work an improved version of the Lagrangian CMC model is proposed to consider interaction by premixed flame propagation along a constant mixture fraction contour between neighboring flame groups.

Low temperature combustion (LTC) may allow simultaneous reduction of nitrogen oxides (NOx) and soot with acceptable compromise in the efficiency of a diesel engine. Recently oxygenate biodiesel fuels were tested to resolve the problem of CO emission at higher exhaust gas recirculation (EGR) rates in LTC operation. In this paper 3-D simulation is performed by KIVA with soybean biodiesel blends of D100 and BD20 for a heavy duty test engine. The oxygen fraction in intake gas is controlled between 7% and 19% to simulate EGR in LTC operation. A surrogate mechanism is constructed by combining the skeletal mechanisms of methyl butanoate (MB) and n-heptane for low and high temperature chemistry. It consists of 76 species and 243 reaction steps with detailed NOx chemistry. The conditional moment closure (CMC) model is employed to address coupling between turbulence and chemistry.

Computation is performed to predict number density, volume fraction and size distribution of soot particles in typical operating conditions of a diesel engine. KIVA has been integrated with the CMC routine to consider turbulence/chemistry coupling and gas phase kinetics for heat release and soot precursors. The compositions of soot precursors are estimated by tracking Lagrangian particles to consider spatial inhomogeneity and differential diffusion in KIVA. The soot simulator SWEEP is employed as a postprocessing step to calculate conditional and integral quantities of soot particles. There are larger particles produced at a higher load or a lower rpm, but no consistent trend for injection timing in the conditional size distribution at the mixture fraction of 0.12. The integral results are obtained for number density, total mass and size distribution by summing up the histories of all tracked particles in the cylinder.

A zero-dimensional code is developed to simulate turbulent spray combustion and NOx and soot emission in direct injection diesel engines. The code consists of two major parts; mixing calculation for the probability density function (PDF) based on the multi-zone model by Hiroyasu et al., (1983) and the flame structure by the conditional moment closure (CMC) model (Klimenko & Bilger, 1999). The skeletal mechanism of n-heptane is employed with the elementary reaction steps for heat release and the NOx chemistry in GRI 3.0. The spray model accounts for evaporation and mixing based on momentum balance of the spray zones, while the CMC model incorporates the conditional flame structures with one fuel group or flame structure for each injection. The spatially integrated density-weighted PDF, F(η), is defined to represent inhomogeneous mixture distribution in the cylinder. The one-equation soot model is employed for prediction of the soot emission.

A low temperature combustion (LTC) diesel engine has been under investigation for reduction of NOx and soot with acceptable compromise in the efficiency through modification of the combustion process. In this paper computational simulation is performed as a preliminary step for development of an LTC diesel engine for off-highway construction vehicles. Validation is performed for major physical models against measurements in LTC conditions. The conditional moment closure (CMC) is employed to address coupling between chemistry and turbulence in KIVA-CMC. The Kelvin-Helmholtz/Rayleigh-Taylor (KH-RT) model is employed for spray breakup and a skeletal n-heptane mechanism for both low and high temperature chemistry. Parametric evaluation is performed for design and operating conditions including EGR rate and injection timing. Results are obtained for efficiency, IMEP, CO, NOx and PM emissions at intake boost pressures of 1, 2 and 3 bar.

This study presents an extended Conditional Moment Closure (CMC) model [1] for turbulent spray combustion of a diesel engine. A spatially integrated CMC equation involves multiple flame structures, evaporation source terms for mixture fraction variance and scaled conditional scalar dissipation rates (CSDR) and probability density functions (PDF). An independent transport equation is solved for each flame group with equal mass of sequentially injected and evaporated fuel group. The implementation strategy in KIVA is to determine mean scalar variables instead of mean reaction rates in terms of conditional flame structures and local PDF's. Simulation is performed to test the suggested CMC model for a heavy duty diesel engine with early injection timings. NOx chemistry [2] is combined with skeletal chemistry of n-heptane [3] while soot is modeled by both one equation model [4] and two equation model [5].

In this study a new model is proposed for turbulent premixed combustion in a spark-ignition engine. An independent transport equation is solved for the mean reaction progress variable in a propagation form in KIVA-3V. An expression for turbulent burning velocity was previously given as a product of turbulent diffusivity in unburned gas, laminar flame speed and maximum flame surface density. The model has similarity with the G equation approach, but originates from zone conditionally averaged formulation for unburned gas. A spark kernel grows initially as a laminar flame and becomes a fully developed turbulent flame brush according to a transition criterion in terms of the kernel size and the integral length scale. Simulation of a homogeneous charge pancake chamber engine showed good agreement with measured flame propagation and pressure trace. The model was also applied against experimental data of Hyundai θ-2.0L SI engine.

Submodels are developed for injection, evaporation and wall impingement of a liquid LPG spray. The injection model determines the quality of fuel as two-phase choke flow at the nozzle exit. Wind tunnel experiments show the spray penetration more sensitive to ambient flow velocity than to injection pressure. Most evaporation occurs during choking, while heat transfer from surrounding air has a negligible effect on downstream droplet sizes. Three dimensional simulation shows that the bathtub cavity is better than the dog-dish cavity for stable flame propagation in lean-burn conditions. The injection timing during the IVC period has a negligible effect, while injection during an intake stroke enhances fuel/air mixing to result in more homogeneous cylinder charge.

Analysis of flow field and charge distribution in a gasoline direct-injection (GDI) engine is performed by a modified version of the KIVA code. A particle-based spray model is proposed to simulate a swirl-type hollow-cone spray in a GDI engine. Spray droplets are assumed to be fully atomized and introduced at the sheet breakup locations as determined by experimental correlations and energy conservation. The effects of the fuel injection parameters such as spray cone angle and ambient pressure are examined for different injectors and injection conditions. Results show reasonable agreement with the measurements for penetration, dispersion, global shape, droplet velocity and size distribution by Phase Doppler Particle Anemometry(PDPA) in a constant-volume chamber. The test engine is a 4-stroke 4-valve optically accessible single-cylinder engine with a pent-roof head and tumble ports.

Combustion characteristics of the emulsified diesel fuels are investigated in a Rapid Compression and Expansion Machine(RCEM). Among the test cases the 40 W/O fuel injected at BTDC 20° has shown the best performance with respect to the efficiency and NOx and soot emissions. The pressure trace of the 40 W/O fuel is characterized by a longer ignition delay and a lower rate of pressure rise in premixed combustion. High speed photographs show reduced flame luminosity and lower flame temperature with the increasing W/O ratio. Micro-explosions of emulsified fuel droplets which affect the local shape and brightness of the flame are identified in magnified flame images.

Recent findings on the characteristics of flame surface density are introduced for turbulent premixed combustion in typical operating conditions of SI engines. The maximum flame surface density tends to show linear dependence on the K -factor defined as a function of the integral length scale and . The flame surface density shows an asymmetric profile in the space with the peak location correlated in terms of the dimensionless parameter, NB, which represents the degree of gradient or counter-gradient diffusion by turbulence. The effects of the K -factor and NB are discussed in the wrinkled flamelet and corrugated flamelet regime respectively. The flame surface density increases at a higher ambient pressure due to decrease in the laminar flame speed and the length scales of flame wrinkling. Comments are made on the turbulent stretch and turbulent flux terms in the Σ -equation in modeling combustion of an SI engine.

The autoignition delay of a turbulent methane jet in a Diesel-like environment is calculated by the conditional moment closure(CMC) model. Methane is injected into hot air in a constant volume chamber under various temperatures and pressures. Detailed chemical reaction mechanisms are implemented with turbulence-chemistry interaction treated by the first order CMC. The CMC model solves the conditional mean species mass fraction and temperature equations with the source term given in terms of the conditional mean quantities. The flow and mixing field are calculated by the transient SIMPLE algorithm with the k -ε model and the assumed beta function pdf. The CMC equations are solved by the fractional step method which sequentially treats the transport and chemical reaction terms in each time step. The predictions in quiescent homogeneous mixture are presented to evaluate the effects of turbulence in jet ignition.

Numerical analysis of the flow field and fuel spray in a gasoline direct-injection (GDI) engine is performed by a modified version of the KIVA code. A simple valve treatment technique is employed to handle multiple moving valves without difficulties in generation of a body-fitted grid. The swirl motion of a hollow-cone spray is simulated by injecting droplets with initial angular momentum around the nozzle periphery. The model for spray-wall impingement is based on single droplet experiments with the droplet behaviors after impingement determined by experimental correlations. Different behaviors of an impinging droplet depend on the wall temperature and the critical temperature of fuel with the fuel film taken into account. The test engine is a 4-stroke 4-valve gasoline engine with a pent-roof head and vertical ports to form a reverse tumble flow during the intake stroke. A hollow-cone spray by a high-pressure swirl injector is employed to enhance mixture preparation and mixing.

Three-dimensional computation of the flow field and fuel spray in a DISC engine is performed using a modified version of KIVA-II. A special valve treatment technique is employed to simulate multiple moving valves without excessive efforts for body-fitted grid generation. The test engine is a 4-valve 4-stroke gasoline engine with a pent-roof head and a hollow-cone spray by a high-pressure swirl injector. The injection strategy is divided into two categories, ‘early’ and ‘late’ injection to optimize the combustion process. A spray-wall impingement model based on a single droplet experiment is implemented to consider both ‘early’ and ‘late’ injection case. Parametric studies are performed with respect to the load, injection timing, duration and position, spark-plug position, and the combustion chamber geometry. Results show that the current numerical analysis is capable of representing the spray motion and mixture formation in an operating engine qualitatively.

Optimization study is performed for the scavenging process as the first step for the development of a poppet-valve type automotive two-stroke diesel engine. The scavenging flow pattern is varied by the RSSV (rotatable shrouded scavenging valve) system, which was designed for application of a shroud valve to an actual engine. The scavenging flow is analyzed by flow visualization and numerical calculations under a steady condition. Water is used as the working fluid, instead of air for effective visualization of the flow pattern in the flow visualization study. More details in the scavenging characteristics are observed by a dye experiment, in which the dye path indicates the flow streamline in the cylinder. In the numerical study, three-dimensional flows are calculated by a modified version of KIVA-2 code, with a special technique to consider the valve and shroud shapes.

A modified version of KIVA-II code is applied to scavenging flow simulation of a four-poppet-valved two-stroke engine. The standard k-ε turbulence model is used with no slip on the wall. Grid is generated through direct interface with three dimensional CAD data using a commercial CAE package. Valve motion is taken into account by identifying the cells occupied by the valve structure and putting them as solid obstacle cells. Results show reasonable trends for variation of the velocity field and fresh air mass fraction distribution with crank angle. Parametric study shows that the intake port orientation does not have much effect on the cylinder flow and scavenging efficiency due to strong flow diversion by the valves.

Multidimensional cycle simulation is performed for a spark ignition engine for the entire four stroke cycle of intake, compression, expansion and exhaust stroke. Combustion is divided into ignition period and flame propagation period. Simple Arrhenius form is used in ignition and Bray-Moss model is used for turbulence controlled flame propagation. Heat flux is calculated by the wall function method in k-ε turbulence model. Grid is generated by a two dimensional CAD package. Results show proper trends for all flow variables, burning rate and heat flux.

A spark ignition engine is simulated for its flow field and combustion using a modified version of KIVA-II. The code handles an arbitrary 3-D computational domain and boundary conditions in the Cartesian coordinate. The valve is fixed at a half open position and the pressure boundary condition is specified at the port upstream. The grid is generated by a 2-D CAD package and later combined to a 3-D. Post-processing is performed by a finite element postprocessor package. Results for a typical spark ignition engine show reasonable trends for the velocity field and flame propagation throughout the intake, compression, combustion and expansion stroke.

Three different models, the law-of-the-wall, a modified law-of-the-wall, and an approximate one-dimensional solution to the energy equation are compared for the spatially-resolved prediction of engine heat tranfer. The multidimensional hydrodynamic code KIVA is used for the fluid mechanic simulation. Two different engine geometries are studied; one being a pancake-shaped chamber, and the other a bowl-in-piston geometry. The comparisons are done for a range of initial conditions of gas flow. Rates-of-pressure-rise were also varied to represent rates typical of those encountered in motored engines, and those found in fired engines. Comparisons with experimental results show that the heat transfer predictions using the law-of-the-wall may be in error when source terms such as the transient, work and chemical energy terms have a significant effect in determining the temperature profile in the boundary layer.