Search Results

The goal of this theoretical-experimental work is to model two-dimensional, unsteady (and steady) two phase flow combustion in internal combustion engines (and steady reactors) and to test and improve the model with parallel experimental programs. The purpose of this research is to make more detailed current understanding of this important family of combustion problems and to aid the development of cleaner and more efficient engines. In this paper, preliminary theoretical-experimental results of our efforts toward the stated goal are presented. The theoretical results are preliminary but prove the feasibility of detailed computations of combustion in internal combustion engines and show how informative such computations can be. In a section of this paper, preliminary results are reported of detailed computations of two-dimensional, unsteady sprays penetrating and vaporizing into an inert gas in a closed volume (without combustion).

Flame fronts were examined in a premixed-charge, spark-ignition, ported engine using a two-dimensional visualization technique with 10 nanoseconds time resolution and 200 microns best spatial resolution. The engine had a pancake chamber, a compression ratio of 8, a TDC swirl number of 4 and was operated at 300 to 3000 rpm with stoichiometric and lean propane/air mixtures. The measurements were made far from, and near to, the cylinder wall. A pulsed laser sheet was passed through the engine and the light scattered by sub-micron TiO2 or ZrO2 seeding particles was collected by a 100 x 100 diode array with fields of view of 1 cm x 1 cm, 2 cm x 2 cm, and 9 cm x 9 cm. The thickness of the flame front is as small as, or smaller than, the 200 micron best resolution of the measurements thus confirming that premixed-charge engine turbulent flames generally appear to be wrinkled laminar flames.

The structure of turbulent flames was examined in a premixed-charge, spark-ignition ported engine using a three-dimensional visualization technique with 10 ns time resolution and 350 µm best spatial resolution. The engine had a pancake chamber, a compression ratio of 8, a TDC swirl number of 4 and was operated at 300, 1200 and 2400 rpm with stoichiometric and lean propane/air mixtures. The second and third harmonic beams of an Nd-YAG laser (532 nm and 355 nm), along with the two strongest beams (first Stokes (683 nm) and first anti-Stokes (436 nm)) from a hydrogen Raman shifter pumped by the second harmonic were used to create four parallel laser sheets each of less than 300 microns thickness. The laser sheets were passed through a transparent quartz ring in the cylinder head parallel to the piston top with vertical separations between successive sheets ranging from 1.5 to 0.9 mm.

Recent measurements by of gas intake flows and exhaust pressure in a motored, ported, single-cylinder engine with strong swirl and roll have been used as boundary conditions to a three-dimensional, transient computer simulation of the flow within the cylinder. For each condition, the calculation is continued over several engine cycles until the periodic solution is obtained. The computed TDC tangential velocity and turbulence intensity are then compared with measured ones. A technique is described to evaluate scavenging efficiency, the fraction of charge that remains in the cylinder over later cycles and the degree of mixedness of fresh and residual charge. For this motored ported engine, it is found that the scavenging efficiency is very low (19.4% at 1200 RPM) and the inflow from the exhaust ports is very significant. For practical ported engines with combustion, the scavenging efficiency is much higher but the inflows from exhaust ports are still expected to be significant.

For practical, extensive 3-D computations for engine improvements, each physical submodel needs to be the simplest that is compatible with the accuracy of all other physical submodels and of the numerics. The addition of one progress variable controlled by one Arrhenius term is shown to be adequate to reproduce Diesel ignition delay in 2-D and 3-D computations. The rest of the model is that used for years by the authors to optimize combustion in reciprocating and rotary engines with premixed and non-premixed charges, including all of its model constants. This minimal Diesel autoignition submodel reproduces well trends and magnitudes of ignition delay versus chamber temperature and pressure. As in experiments, it is found that multiple ignition sources develop in rapid succession at various locations around the fuel spray after the first ignition event.

The exciplex fluorescence technique has been used to separately visualize liquid and vapor phase fuel in engines since its development by Melton. However, as a fluorescence technique it has the potential to be quantitative and the underlying assumptions have been outlined by Melton. An initial quantitative application of the TMPD/naphthalene system, based on these assumptions, applied to a hollow-cone spray in a two-stroke engine, indicated that it substantially over-estimates the concentration of fuel vapor about TDC. The reasons for the discrepancy were investigated and it was concluded that a major factor is the effect of temperature on the photophysics of the species involved. Thus the absorption spectra of the exciplex dopants were determined at temperatures up to 700 K. These experiments showed that the increase in absorption with temperature above 500 K is responsible for the failure of the earlier calibration.

Two-Dimensional, single-shot spontaneous Raman measurements of methane concentration were performed in an optically accessible engine after direct injection with the use of modified air-assisted injector. The spatial resolution of the measurements was determined by the thickness of the laser sheet which was 0.8 mm. The error in the methane number density measurement was determined by the noise in the intensified camera output and was 16% of pure methane number density at the experimental conditions. Effective suppression of the stray light background was the main experimental difficulty. Satisfactory results were acquired only when the spark plug was substituted by a plug covered with a velvet-like, black piece of cloth. These preliminary results show that, for the specific engine configuration, fast mixing of the charge yields a very mild stratification after the end of injection.

Atomization and full-cone sprays from single cylindrical orifices are considered. The following subjects are reviewed: the structure of the breakup region; the structure of the far field; modern models that, given the outcome of the breakup process, compute the steady and transient of sprays; some comparisons with detailed measurements; and some practical applications. The following conclusions are reached: the spray breakup and the development regions are the most relevant in engine applications; the inner structure of the breakup region is still largely unknown; two- and three-dimensional spray models are available but remain mostly untested, particularly in their vaporization and combustion components, in part because of a lack of accurate measurements in controlled engine-type environments; engine applications of such models are, nonetheless, recommended for very valuable learning, interpretative, and exploratory studies, but not for predictions.

Planar instantaneous fuel concentration measurements were made by laser-induced fluorescence of 3-pentanone in the spark gap just prior to ignition in a direct-injection spark-ignition engine operating at a light load, highly stratified condition. The distribution of the average equivalence ratio in a circle of 1.9 mm diameter centered on the spark plug showed that a large fraction of the cycles had an equivalence ratio below the lean limit, yet acceptable combustion was achieved in those cycles. Further, weak correlation was found between the local average equivalence ratio near the spark plug and the time required to achieved a 100 kPa pressure rise above the motoring pressure, as well as other parameters which characterize the early stages of combustion. The cause for this behavior is assessed to be mixture motion during the spark discharge which continually convects fresh mixture through the spark gap during breakdown.

Results of experiments performed on a direct-injection two-stroke engine using an air-assisted injector are presented. Pressure measurements in both the engine cylinder and injector body coupled with backlit photographs of the spray provide a qualitative understanding of the spray dynamics from the oscillating poppet system. The temporal evolution of the spatial distribution of both liquid and vapor fuel were measured within the cylinder using the Exciplex technique with a new dopant which is suitable for tracing gasoline. However, a temperature dependence of the vapor phase fluorescence was found that limits the direct quantitative interpretation of the images. Investigation of a number of realizations of the vapor field at a time typical of ignition for a stratified-charge engine shows a high degree of cycle to cycle variability with some cycles exhibiting a high level of charge stratification.

Results are presented of 3-D computations of direct injection of gaseous methane and of liquid tetradecane through a multi-hole injector into a Diesel engine. The study focusses on the distribution of fuel/air ratio within the resulting gas and spray jets under typical Diesel conditions prior to ignition. It is shown that for a significant time after start of injection, the fraction of the vapor fuel which is in richer-than-flammable mixtures is greater in gas jets than in sprays. For methane injection, it is also shown that changing some of the flow conditions in the engine or going to a poppet-type injector, does not result in improved mixing. An explanation of these results is provided also through an analysis of the self-similar gas jet and 2-D computations of gas and spray jets into constant pressure gas. A scaling for time and axial distance in the self-similar gas jet also clarifies the results.

Combustion in a divided-chamber, stratified-charge engine is considered and flame and pressure results obtained with a two-dimensional, unsteady model are compared with corresponding engine data. The model is applied to eight engine conditions differing in speed, load and size of the prechamber orifice. The model employs one overall chemical reaction rate, the k-ε representation of turbulence, and a wall heat loss proportional to the heat release. The computed results are shown to be in good agreement with the experimental ones in spite of the complexity of the problem and the early stages of detailed model validation studies. They are also shown to compare somewhat better than earlier ones obtained with an ad-hoc jet turbulence model. Both studies prove the importance of the prechamber jet to the overall combustion process for the particular engine investigated.

A three-dimensional model for flows and combustion in reciprocating and rotary engines is applied to a direct-injection stratified-charge rotary engine to identify the main parameters that control its burning rate. It is concluded that the orientation of the six sprays of the main injector with respect to the air stream is important to enhance vaporization and the production of flammable mixture. In particular, no spray should be in the wake of any other spray. It was predicted that if such a condition is respected, the indicated efficiency would increase by some 6% at higher loads and 2% at lower loads. The computations led to the design of a new injector tip that has since yielded slightly better efficiency gains than predicted.

Simultaneous fuel distribution images (by shadowgraph and laser-induced fluorescence) and cylinder pressure measurements are reported for a combusting stratified-charge engine with a square cup in the head at 800 RPM and light load for two spark locations with and without swirl. Air-assisted direct-injection occurred from 130°-150° after bottom dead center (ABDC) and ignition was at 148° ABDC. The engine is ported and injection and combustion take place every 6th cycle. The complicated interaction of the squish, fuel/air jet, square cup, spark plug geometry and weak tumble gives rise to a weak crossflow toward the intake side of the engine with no swirl, but toward the exhaust side in the presence of strong swirl, skewing the spray slightly to that side.

In a previous study, the authors compared the fuel-air mixing characteristics of gas jets and sprays in Diesel engine environments in the absence of combustion. A three-dimensional model for flows and sprays was used. It was shown that mixing was slower in gas jets relative to fast-evaporating sprays. In this study, which is an extension of the previous one, the direct-injection of gasesous methane, gaseous tetradecane and liquid tetradecane are studied using the same three-dimensional model. This study concentrates on combustion. It is shown that the fuel-air mixing rate and hence the burning rate are initially slower with gas injection.

Laser Doppler velocimetry has been used to make cycle-resolved velocity and turbulence measurements in a homogeneous-charge, spark-ignition engine. The engine had a ported intake and disc-shaped chamber with a compression ratio of 8 to 1. It was operated at a speed of 1200 rpm and with a TDC swirl number of 4. A stoichiometric propane-air mixture was used, and ignition was near the wall. The velocity measurements were made at three spatial locations at the midpoint of the clearance height. Tests were made to determine whether the presence of the flame affected the accuracy of the velocity measurements. It was found that the ensemble-averaged mean velocity shows a small deviation, and the rms fluctuation intensity is significantly influenced, but the effects appear to be confined to the flame zone. Data rates were sufficiently high in the preflame and postflame regions to determine the velocity history in each cycle (cycle resolved).

Laser Doppler velocimetry has been used to make cycle-resolved velocity and turbulence measurements in a homogeneous-charge, spark-ignition engine. The engine had a ported intake and disc-shaped chamber with a compression ratio of 7.5 to 1. It was operated at a speed of 1200 rpm and with a TDC swirl number of 4. A stoichiometric propane-air mixture was used, and ignition was near the wall. Measurements of the tangential velocity component were made in both firing and non-firing cycles at nine spatial locations along a radius 180 degrees downstream of the spark. The radial velocity component was also measured at four of the locations. All measurements were made in the center of the clearance height. Tangential component measurements were made as close as 0.5mm from the cylinder wall, and the radial component was measured as close as 1.5mm from the wall.

Computations are reported of transient axisymmetric pulsating and evaporating sprays that account also for drop collisions and coalescence. It is found that, for the same upstream and gas conditions, pulsating injections result in smaller drops than continuous injections. The difference is particularly marked at high gas densities and is due to the inhibition of collisions and coalesce of drops generated by the gas gap in between the pulses. However, the tip penetration rates are not markedly different for continuous and pulsating injections. For transient evaporating sprays it is found that all drops except the largest evaporate within a well defined distance from the injector. Beyond this distance only vaporized liquid and entrained gas continue the penetration. For engine applications the length of the liquid core is found to be of the order of centimeters and sensitive to conditions. In particular it decreases with increasing injection pressure, gas temperature, and gas density.

The combustion chamber pressure computed with a three-dimensional model is compared with the measured one in a rotary engine fueled with mixtures of natural gas and air. The rotary engine has a rotor displacement of 654 cm3, a compression ratio of 9.4 and uses 2 ignition sparks. The model incorporates a k-ϵ submodel for turbulence, wall function submodels for turbulent wall boundary layer transport, and a hybrid laminar/mixing controlled submodel for species conversion and energy release. Nine cases are considered that cover a wide range of engine operating conditions: rpm of 2503-5798, volumetric efficiency of 35.7-100.5% and equivalence ratio of 0.59-1.15. In all cases the computed and measured pressures agree within 12%.

Results are presented of two-dimensional computations of air-assist fuel injection into engines with bowl-in-piston and bowl-in-head, with and without swirl and for early and late injection but without combustion. The general finding is that swirl tends to destroy the head vortex of the air/fuel jet and results in a faster collapse of the spray cone toward its axis. Faster collapse is also promoted by high density of the chamber gas (e.g. late injection) and bowl-in-head design (limited availability of chamber gas around the spray, presence of walls and delayed influence of squish by the injector). With enhanced collapse, fuel-rich regions are formed around the axis and away from the injector. With reduced collapse, the radial distribution of the fuel is more uniform. Thus swirl tends to lead to both slower vaporization and richer vapor mixtures. Also, with strong swirl the rich mixtures tend to end up by the injector; without swirl, by the piston.