Search Results

Two distinct atomization strategies are contrasted through the measurement of time and spatially dependent mass flux. The two systems investigated include a pressure atomizer (6.9 MPa opening pressure) and an air-assist atomizer. Both systems have potential for use in direct injection spark ignition engines. The mass flux data presented were obtained using a spray patternator that was developed to allow phased sampling of the spray. The temporal mass related history of the spray was reconstructed as volume versus time plots and interpolated mass flux contour plots. Results indicate substantial differences in the distribution of both mass and mass flux in space and time for the two injection systems. For example, the pressure atomizer at high mass delivery rates produced a spray that collapsed into a dispersed cylindrical shape while at low rates, generated a hollow cone structure.

This paper reports on the investigation of injection pressure upon the droplet behavior in transient diesel sprays. Phase/Doppler results for a Diesel spray with a maximum fuel injection line pressure of 105 MPa are compared with previously acquired droplet size and velocity measurements for a Diesel spray with an injection pressure of 21 MPa. All measurements reported here were made in atmospheric conditions at a position near the nozzle. It is shown in these results that the droplet velocity and size profiles do maintain similarity despite the substantial change in injection pressure. Specific characteristics, for example, the appearance of subtle waves in the time-dependent spray data, are present in both data sets. Comparison of the measured droplet velocities and diameters with Weber number based stability criteria shows that increased injection pressure produces a higher percentage of droplets that are likely to breakup.

Previous experiments using an air-assisted spray in a two-stroke direct-injected engine demonstrated a significant improvement in combustion stability at part-load conditions when a wide injection spray was used. It was hypothesized that the decrease in variability was due to the spray following the combustion chamber wall, making it less affected by variations in the in-cylinder gas flows. For this study, experiments were conducted to investigate engine spray combustion for cases where engine performance was not dominated by cyclic variation. Combustion and emission performance data was collected for a wide range of injection timings at several speed/load conditions. Experimental data for combustion shows that combustion stability is relatively unaffected by injection timing changes over a 40 to 100 degree window, and tolerant to spark gap projections over a range of 0.7 to 5.2 mm, depending on operating conditions.

Laser Doppler velocimeter results are presented for the mean velocity and turbulence intensity measured in a motored research engine. The compression of complex bulk motions created during induction produces turbulence as the piston approaches top dead center. The turbulence field is shown to be isotropic but nonhomogeneous. A zero-dimensional computer simulation based on an averaged k-ϵ model is shown to adequately predict the decay of turbulence at a point in the flow after the production phase is completed. Cylinder pressure measurements were recorded for homogeneous stoichiometric combustion for a range of engine speeds and ignition locations. A two-zone (burned and unburned gases) thermodynamic model accurately predicts the measured pressure histories when the turbulence results determined from the motored tests are used to establish initial conditions for the combustion model.

Laser Doppler velocimeter results are presented for the mean velocity and turbulence intensity measured during combustion in a research engine. Simultaneously with each LDV measurement, the cylinder pressure and gas state (unburned or burned) were measured so that conditional sampling techniques could be used in the data-averaging procedure. Measurements of the mean velocity component in the direction of flame propagation agree well with a computer simulation of the induced velocities generated by the volume expansion of the burned gases. Mean velocities measured parallel to the flame surface are shown to be complex because a small amount of swirl was present. Conditional sampling on the time of flame arrival at the LDV probe volume revealed a thirty percent cyclic-variation bias error in the turbulence component normal to the flame.

Measurements are presented for the turbulence intensities and mean velocities obtained in a research engine in which a grid was used to create a flow field characterized by negligible mean motions and homogeneous and isotropic turbulence at the time of ignition. Pressure measurements for homogeneous stoichiometric combustion indicate a very low level of cyclic variation. The combustion-induced mean flow field is shown to be characteristic of a one-dimensional compression of the unburned gases, which produces a small increase in the bulk turbulent kinetic energy ahead of the flame. Most of the effect of combustion appears to occur locally, as the turbulence in the preflame gases close to the flame front is strongly amplified in the direction of flame propagation. Parallel to the flame surface there is little effect until the flame has propagated nearly all the way across the chamber.

Cyclic variation is examined by: (1) conditional grouping and heat-release analysis to reveal different modes of combustion, (2) considering the order in which the burn modes occur to establish prior-cycle effects and (3) comparing the measured variation in IMEP with data generated by simple models. Results show that several burn modes may exist, particularly under fuel-lean conditions. Prior-cycle effects also become more obvious as the air-fuel ratio is increased. Finally, comparisons with data generated by simple models show that the nature of cyclic variation may range from completely stochastic to a superposition of a non-chaotic deterministic process on a stochastic process.

An in-cylinder optical sensor has been developed and tested for use in spark-ignition engine combustion analysis and control, This sensor measures the luminous emission in the near infrared region. Results of these tests show good correlation between the measured luminosity and traditional combustion parameters, such as location and magnitude of maximum cylinder pressure, and location and magnitude of maximum heat release. Engine performance indicators, such as the indicated mean effective pressure (IMEP), also can be determined accurately with the measured luminosity combined with other engine operating parameters, e.g. intake manifold pressure. In-cylinder air-fuel ratio can be determined with accuracy over an ensemble of 100 cycles.

In the first part of this study, a one-dimensional code was used to compare predictions from six different two-equation turbulence models. It is shown that the application of the traditional k-ε models to the viscous-dominated region of the boundary layer can produce errors in both the calculated heat flux and surface friction. A low-Reynolds-number model does not appear to predict similar non-physical effects. A new one-dimensional model, which includes the effect of compression, has been formulated by multiparameter fit to the numerical solution of the energy equation. This model can be used in place of the law-of-the-wall to calculate the surface heat flux. The experiments were performed in a specially-instrumented engine, allowing optical access to the clearance volume. Measurements of heat flux, swirl velocities, and momentum boundary layer thickness were made for different engine speeds.

Conditional sampling of cylinder-pressure data is used to investigate cyclic variability in a premixed-charge spark-ignited engine operating under fuel-lean conditions. Unlike straight ensemble averaging of pressure data, conditional sampling applies a set of constraints to the pressure data such that like combustion events can be identified and grouped together. Ensemble averaging of pressure data from an engine that exhibits significant cycle-to-cycle variation is shown to produce a mean pressure history that is not representative of the combustion process. Conditional sampling provides a means of identifying and analyzing the different groups of pressure histories and therefore the different types of combustion processes that occur in an engine that exhibits cyclic variability.

Predictions of the volume flow rate through an inlet port were produced by four different commercially available CFD programs suitable for use in a steady flow simulation. These predictions were compared with experimental measurements of an inlet port's discharge coefficients. The experiment performed was a typical steady state flow bench test for an inlet port. Volume flow rates were measured at five different valve lifts. The largest valve lift tested (12.24mm) was the maximum value of lift under actual operation. The smallest valve lift was typical of early valve opening. The tests were performed at two different pressure differences across the inlet port and valve at each of the five different valve lifts. All predictions were made using an RNG k-ε turbulence model. Standard wall functions were used to predict wall friction effects and the energy equation was included to account for compressibility effects.

Gas-phase in-cylinder mixing was examined by two different methods. The first method for observing mixing involved planar Mie scattering measurements of the instantaneous number density of silicon oil droplets which were introduced to the in-cylinder flow. The local value of the number density was assumed to be representative of the local gas concentration. Because the objective was to observe the rate in which gas concentration gradients change, to provide gradients in number density, droplets were admitted into the engine through only one of the two intake ports. Air only flowed through the other port. Three different techniques were used in analyzing the droplet images to determine the spatially dependent particle number density. Direct counting, a filtering technique, and autocorrelation were used and compared. Further, numerical experiments were performed with the autocorrelation method to check its effectiveness for determination of particle number density.

In-cylinder heat flux and temperature measurements were obtained in an air-cooled four-stroke utility engine for a range of air-fuel ratios. For these measurements, the magnitude of the integrated heat flux peaked at the stoichiometric air-fuel ratio, with an approximately linear decrease on either side of stoichiometric. Advancing the spark generally increased the magnitude of the integrated heat flux. Evaluation of the Brake Specific Integrated Heat Flux (BSIHF) mitigated these trends, and, the effects of changes in timing were eliminated for some operating conditions Examination of the BSIHF from the compression and expansion stroke showed behavior mimicking the full cycle BSIHF. However, the fraction of the total flux contributed by this portion of the cycle varied greatly from approximately 98% of the total to approximately 75% of the total.

Measurements of the instantaneous heat flux at three positions on the cylinder head surface, and the steady-state cylinder head temperatures at four positions on the cylinder head have been obtained. Engine tests were performed for a range of air-fuel ratios including regimes rich of stoichiometric, stoichiometric, and lean of stoichiometric. In addition, ignition timing was advanced in increments from 22° BTDC to 40° BTDC. All tests were run with the throttle either fixed in the wide open position, or fixed in a position that produced 75% of the maximum power with the standard ignition timing and an air-fuel ratio of 13.5. This was done to ensure that changes in air mass flow rate were not influencing the results. In addition, all tests were performed with a fuel mixture preparation being provided by system designed to deliver a homogeneous premixed charge to the inlet port. This was done to ensure that mixture preparation issues were not confounding the results.

Three different carburetor types have been tested to observe differences in the characteristics of the fuel/air mixtures produced. To characterize the fuel/air mixtures, two diagnostics have been applied: 1) High speed movies and subsequent analysis of the exit flow, and 2) measurement of the A/F ratio found in different positions within the intake manifold. The three different carburetor types that have been studied include a fixed-venturi, fixed-jet butterfly carburetor, a slide-valve carburetor, and a constant-velocity carburetor. Each carburetor type produced a unique set of exit flow characteristics, with differences in the optical density of fuel exiting the carburetor, and differences in the apparent amount of fuel on the intake manifold wall, entrained in the air flow, and in vapor phase.