Compression ignition combustion of a premixed mixture is attracting attention of the combustion engineers as a promising system for low emission and high efficiency engines. In this paper, a historical review is presented about the development of compression ignition combustion systems of premixed mixtures at ACE & New ACE, then, a future perspective of the system is reviewed putting some emphasis on fuel injection systems. As a conclusion, a variable orifice nozzle will have better potential for proper fuel distribution in the combustion space in a wide range of the engine operation in case of this combustion system. Fuel properties have important roll in this system, and combined with some exhaust treatment system, compression ignition combustion of premixed mixtures will make a fuel efficient and clean engines for the next generation.

View full abstract

Since 1988, particulate matter emission regulations in the US for heavy-duty diesel engines have mandated a reduction from 0.6 g/bhp-hr to the current level of 0.1 g/bhp-hr. As large an improvement as this has been, however, looming in the not-to-distant future is a requirement for an additional order-of-magnitude reduction, to 0.01 g/bhp-hr by 2007, as illustrated in Fig. 1. It will take a major effort by industry to reach this target, and it will most likely require the use of a particulate trap. But this large reduction in total particulate mass will also create a new problem - how to measure it. The current gravimetric procedure of weighing a sample collected on filter paper will be impractical because of the long time required to collect sufficient mass to be detectable. This problem of measurement sensitivity is compounded by the fact that the size of particles emitted by contemporary engines is far smaller than that of engines of 1988. This is why particulate matter emissions are no longer visible. However, achievement of this reduction in size came at the cost of a tradeoff with the number of particles emitted, which has increased by several orders of magnitude and poses a potential health concern. These new issues of size and number may prove to be as important, or even more so, than particulate mass, raising questions about whether "what" is regulated may also change in the future. Compounding this problem are newly raised issues regarding whether the nanoparticles observed in a dilution tunnel are representative of tailpipe exhaust dilution by the atmosphere. In order for regulators to address these issues, improved measurement techniques are needed now, to provide a better understanding of the importance of size and number on environmental and health issues. It is also important to note that beginning in 2004, gasoline fueled vehicles will be required meet the same regulations as light-duty diesel vehicles. Current port fuel-injected gasoline engines will have no difficulty meeting the 2004 levels, but it is much less certain for gasoline direct-injection engines, or for either type in 2007. Gasoline engines, in general, are known to emit particulates during cold start, and gasoline direct-injection engines have been shown to produce measurable PM during lean burn operation. Industry will also need new diagnostic tools to help them meet the 2007 particulate matter requirements. As engine emissions continue to become cleaner due to improvements in the combustion process, the contribution from engine transients will play an everincreasing role. Only a few of the particulate measurement instruments currently in use respond in real time, and it is doubtful that these will have the sensitivity required for the new regulations. In this paper, I review the diagnostic tools for particulate matter measurement that are currently available commercially, looking first at those that measure total mass, volume, area, or number, and then those that can characterize particles based on size. I next describe some new techniques that are currently only being used in research, and some that have yet to be demonstrated. I conclude with my recommendations for the instruments most suitable for use today, and my projections for the new techniques that show the most promise.

View full abstract

The global energy and fuel needs now and even in the foreseeable future will still be met by fossil primary energy sources. It is expected that cheap sources might be depleted in 30 to 50 years from now, starting with mineral oil to be followed by natural gas. This should cause increases in fuel prices and a stronger political and economical dependence on countries providing primary energies. Nuclear energy and renewable energies might not yet be available by then in sufficient amounts to relieve this situation significantly. The world-wide efforts to develop alternative power trains, e. g. electric drives with batteries or fuel cells, will provide locally emission-free propulsion systems. Their contribution to reducing fuel consumption and CO_2 emissions, however, will depend also on the availability of renewable fuels. Over an extended transition time synthetic fuels made from natural gas (and later even from coal) by well-known technologies will acquire an increasing market share. Their production economy will improve with rising fuel prices. The quality of synthetic fuels is superior to today's fuels opening up new avenues for engineering internal combustion engines with still better fuel economy and minimum emissions.

View full abstract

The science of optics has always played an important role in the measurement and understanding of combustion phenomena, including not only laboratory flames but also practical combustion devices such as the internal combustion engines. In particular, being the combustion in direct injection systems a two-phase, turbulent mixing-controlled process that includes short time scale phenomena such as turbulence production and dissipation, spray breakup and evaporation, and pollutants formation it is appropriate to make investigation by non intrusive diagnostic techniques for their intrinsic high temporal and spatial resolution. The paper reviews the optical techniques currently under application to investigate the in-cylinder fluid dynamics, combustion, pollutant formation processes as well as to characterize the exhaust emissions of current internal combustion engines. The review starts with the laser Doppler based techniques for measuring the fluid dynamic field both on air flow test rig and in engines operating under motored conditions as well as the droplet size and velocity of high pressure jet for direct injection engines. Then, it proceeds discussing the potential of polychromatic light scattering, extinction and absorption techniques (Fig. 1) as diagnostic tool for engine investigation alternative to laser induced incandescence (LII) and laser induced fluorescence (LIF). In the second part, polychromatic light extinction and absorption for liquid and vapor distribution measurements in an engine with large optical access is discussed. In the third part, simultaneous extinction, scattering and flame chemiluminescence measurements from ultraviolet (UV) to visible for obtaining detailed information about combustion precursors species during early soot formation and chemical properties of nanometric carbonaceous particles are introduced. In the last part, the potential of broadband (190-500 nm) extinction and scattering spectroscopy is demonstrated to detect and evaluate in real time the size and the number concentration of the exhaust soot particles.

View full abstract

Current emission legislation forces the automotive industry to significantly reduce the exhaust gas emissions of passenger cars. Especially the contributions of soot and nitrogen oxide of diesel engines will be the main problems in the future. Here, exhaust gas aftertreatment systems could be a possible solution but since modern injection systems-like common rail-deliver more degrees of freedom referring to the injection process, also the exhaust gas reduction by optimisation of the combustion process has a high potential. In this study, a passenger car common rail system applied to an optically accessible one-cylinder transparent engine (AUDI V6 TDI) was investigated by simultaneously detecting spray and flame propagation. When using a pre-injection, the following main injection ignites almost without any ignition delay. Therefore a spatial coexistence of liquid phase and flame can be observed over a large part of the engine cycle. The lack of oxygen in the areas where the flame can be detected enlarges the danger of higher soot formation because the quality of fuel evaporation and mixture formation is reduced. The comparison of a conventional common rail injector to a piezo-driven injector at identical rail pressures showed advantages for the piezo-driven system. The fuel intake is much faster but the tendency to develop a wall film is not higher. The reasons for this can be found in stronger wave structures of the injected fuel which is an assumption for better air entrainment and vaporisation of the fuel. Careful dimensioning of those systems might completely avoid a wall film in the combustion bowl.

View full abstract

In CIMAC in May 2001, the largest congress for marine and stationary engines in the world, many engine builders have announced that they apply the water injection into the cylinder to reduce the NOx emission. There are three methods of water injection into the cylinder, FWE (Fuel Water Emulsion), SFWI (Stratified Fuel Water Injection) and DWI (Direct Water Injection) [1]. At the last COMODIA in 1998, the authors presented about the effect of stratified fuel water injection to reduce NOx and smoke at the same time [2]. In the present study, to confirm the effect of fuel-water emulsion on combustion, experiments have been carried out using a visual engine and a visual combustion chamber. Figure 1 shows the burning flames in the two cases, (a) pure BFO (Bunker Fuel Oil) and (b) BFO-water emulsion. The combustion system of the visual engine simulates the one for low-speed marine engines. According to it, the flame of (b) shows the better combustion state, less soot-cloud and less after-burning than the flame of (a). In the presentation, reduction of the flame temperature measured by the two-color method and reduction of the soot formation inside the flame observed by the back diffused laser photo technique applying the fuel-water emulsion are introduced.

View full abstract

Medium and large-sized diesel engines have two economic advantages : high efficiency and use of cheap fuel, the latter being made possible by the use of heavy fuel. But recently, the quality of heavy fuel is becoming worse and utilization is becoming more difficult. Main problems of such heavy fuel are long ignition delay and long after-burning. It can easily be supposed that the former is depending on the first injected fuel and the latter on the fuel injected last. In this paper a new fuel system called Stratified Fuel Injection System is proposed as shown in Fig. A1. This system makes it possible to inject a small amount of light fuel at the beginning and at the end of injection duration. Heavy fuel is injected in between and sandwiched by the light fuel. A prototype of that injection system is assembled and applied to a high-speed engine. As a result, thanks to only 10% addition of light fuel, the heavy fuel can be burned well just the same as pure light fuel as shown in Fig. A2.

View full abstract

Reduction of the Particulate Matter (PM) is very important to achieve the clean diesel engines. Especially, reduction of the Soluble Organic Fractions (SOF) are very important, because the particulate trap can eliminate primarily the Insoluble Fractions (ISF) but does not effectively reduce the SOF, while the latter contains the harmful components for human. A clarification of SOF production in the combustion chamber is required for the SOF reduction. This paper is concerned with the formation of Particulate Matter (PM) in direct injection (DI) diesel engines. A system featuring an electromagnetically actuated sampling valve with internal N_2 dilution was developed for sampling of PM directly from the combustion chamber. The concentration of Total Particulate Matter (TPM), SOF and ISF, were measured at the twenty different locations in the combustion chamber at different sample timings (different crank angles). The time resolution of the sampling valve is dominated by the opening duration of the valve, which is an order of 1 or 2 milli seconds. In this study, apparent time resolution was improved by the following measure. The sample timing was determined so that the neighboring sample timings were included within the sampling duration. The measured value at each sample timing could be recognized as an averaged value of the instantaneous value at the sample timing and at the neighboring timings. Instantaneous values can be obtained by solving the simultaneous equations between average and instantaneous values. The weight factor were selected to obtain suitable corrected values by trial and error. The combustion gases (CO, CO_2, O_2, THC) were also analyzed by the gas chromatography for the local air fuel ratio calculation. The concentration of SOF was higher at the sampling positions on spray flame axis. The concentration of ISF was higher at the sampling positions on downstream of the spray flame. Both SOF and ISF concentration are higher at near the wall than away from the wall. It is also made clear that the PM formation is strongly affected by the wall quenching at the combustion chamber wall.

View full abstract

The chemical gas-phase reduction process used to reduce nitric oxide (NO) in diesel engine exhaust is applied to a high-speed, light-duty diesel engine. The chemical gas-phase reduction process involves adding a methylamine (CH_3NH_2) in water solution to the exhaust gas as a NO reduction agent. In our previous experiments, the NO reduction process was examined by passing a fraction of the diesel exhaust through a heated quartz flow reactor. The results revealed that the sufficient mixing of methylamine with diesel exhaust effectively breaks down NO into nitrogen and water. It was also found that particulate matter such as soot was found to inhibit the NO reduction. When the particulate matter was filtered from the exhaust, 80% NOx removal was achieved at a reactor temperature of 420 ℃ and a residence time of about 0.14 seconds with a molar ratio CH_3NH_2/NO (φM) of 1. In this study, an experimental selective non-catalytic NO reduction system designed to be used with a diesel engine is used to evaluate this technique for practical use. Experiments were conducted using a single-cylinder, four-cycle, water-cooled, direct-injection, 857 cm^3 of stroke volume, naturally- aspirated diesel engine at an engine speed of 1800 rpm and a brake mean effective pressure of 0.63 MPa. The NO concentration at the inlet of the mixing chamber was 1250 ppm. The same molar amount of methylamine as NO in the exhaust was added to a mixing chamber. Two different mixing chambers with different volumes and residence times (0.1 s and 0.17 s) were tested. The temperature at the inlet of the mixing chamber (T_<inlet>) was approximately 420 ℃. Figure 2 shows the NOx reduction ratio (R_<NOx>) of methylamine processes with and without the installation of a particulate filter. In this figure, the results of the flow reactor experiment from the previous study are shown for comparison. Longer residence times were required to achieve a given level of NOx reduction in unfiltered exhaust, suggesting that the presence of particulate matter inhibits NO reduction. For the standard residence times, the process achieved 64% NO reduction in unfiltered diesel exhaust, increasing to 80% NO reduction when a particle filter is fitted to the system.

View full abstract

Recently, an on-board measurement system that is capable of measuring real-time mass emission of nitrogen oxides (NOx), fuel consumption, road load, and engine output simultaneously has been developed. The system consists of a data recorder and a variety of sensors including an air-to-fuel ratio sensor and a NOx sensor. The system can be placed on the passenger seat and operated without external power. Test results of NOx mass emission and fuel consumption obtained by on-road measurements of diesel vehicles have been already reported. In the present investigation, a total hydrocarbon (THC) analyzer using flame ionization detector (FID) and a smoke-meter using opacity method are added to the on-board measurement system, because THC and particulate matters (PM) are paid much attention as well as NOx. The on-board measurement system was installed in a diesel vehicle and measurements were taken on a chassis dynamometer and on public road. As a result, it can be shown that the on-board measurement system can measure mass emission of THC as well as NOx or other items. Mass emission of PM is also possible to be obtained from smoke-meter output, by applying a calibration line that is pre-determined by comparing with conventional filtering method. Evaluation of the results, using a chassis dynamometer and CVS-tunnel system as a reference, reveals fair correlation : from -3 % to 8 % for THC mass emission, and from -10 % to 12 % for PM mass emission. At the on-road test, it has been observed that the patterns of THC and PM mass emission during on-road runs show good agreement with those of NOx mass emission and fuel consumption. And the effects of the EGR function for NOx and PM mass emission have been clearly observed by the on-board measurement system.

View full abstract

Aiming at higher thermal efficiency, various types of spray/jet combustion have been developed for internal combustion engines such as diesel engines, gasoline direct-injection engines, gas-fueled direct-injection engines and premixed compression ignition engines. In the combustion techniques employed in these engines, there is much difference not only in fuel composition, charge air pressure and temperature but also in heterogeneity of fuel-air mixture, i. e., the extent of fuel-air mixing. The difference decides the characteristics of ignition and combustion processes in each engine. Therefore, the demand for better understanding of effects of fuel-air mixing on the ignition process has grown. From this point of view, in this study, the modeling of spray ignition processes was carried out with the emphasis on fuel-air mixing process. In the developed model, a stochastic mixing model and a reduced chemical reaction model are combined to represent the simultaneous progress of fuel-air mixing and chemical reaction. By using this model, effects of mixture heterogeneity on spray ignition processes were investigated. First, trends and effects of heterogeneity in temperature and equivalence ratio of mixture were investigated. As shown in Fig. A-1, ignition delays of heterogeneous mixture (spray ignition) have similar temperature dependence to those of homogeneous mixture at low temperatures, while at high temperatures the heterogeneous delays scarcely reflect the ignition characteristics in the homogeneous case. One of the reasons is the difference of mixture heterogeneity with temperature as shown in Fig. A-2. It is suggested that the combined effect of this heterogeneity and existence of the most ignitable mixture at each temperature of surrounding air is responsible for the phenomena. Investigations were also carried out on effects of turbulence scale and of air entrain rate on temperature dependency of ignition delay. The effects of heat absorption rate and distribution of equivalence ratio explain the phenomena caused by the change of these characteristics. Finally, the effects of nozzle orifice size and injection velocity were investigated as an example of simultaneous change of mixing rate and flow rate of entrained air.

View full abstract

It is reported in [1] that in n-heptane spray auto-ignition at constant volume under Diesel-like conditions is substantially affected by the pre-turbulence level. In these experiments, different levels of initial turbulent kinetic energy were generated using a multi hole plate that was moved through the combustion chamber. The turbulence generator was calibrated by LDA. The ignition delays were determined by studying the light emissions in the UV and visible wave length range by optical fibers and a photomultiplier. It was also possible to modify the pre-ignition level of turbulence by the injected spray especially in the regime of pilot injection. To evaluate these effects, a numerical study was done using the detailed chemistry approach incorporated into the KIVA-3 spray combustion code. The basic novelty of the proposed methodology, see [2, 3], is the application of a generalized partially stirred reactor, PaSR, model, to treat detailed oxidation kinetics of hydrocarbon fuels assuming that chemical processes proceed in two successive steps : the reaction act follows micro-mixing simulated on a sub-grid scale. If the all Re number RNG κ-ε model is employed, the micro-mixing time can be consistently defined giving the combustion model in a "well-closed" form. The detailed mechanism integrating the skeletal n-heptane oxidation chemistry with the kinetics of aromatics (up to four aromatic rings) formation for rich acetylene flames [4], consisting of 117 species and 602 reactions, was validated in numerical kinetic analysis, and the reduced mechanism (60 species, including soot forming agents up to the third aromatic ring and NOx species, 237 reactions) was used in the spray combustion simulations. The model application was illustrated by comparison of predicted and measured ignition delays at constant volume at different levels of preturbulence, and good agreement was found (see Fig. 1). Different definitions for the ignition delay were compared and the best agreement was achieved, when the start of ignition was defined as the moment when the average temperature in the combustion chamber exceeds the initial temperature by 1 %. The moderate increase in the level of pre-turbulence leads to a reduction in ignition delay owing to more rapid mixing in forming the ignitable mixture. This effect was also pronounced when the "pilot" injection with different injection schedules was analyzed.

View full abstract

The aim of this paper is to study numerically the detailed flame zone structure of DI diesel sprays during combustion at engine-like conditions. To address this issue, the KIVA-3 code was modified to include complex chemical mechanisms. A "subgrid", partially stirred reactor model was applied to handle turbulence-chemistry interaction. Diesel fuel is assumed to be single-component and its oxidation chemistry is represented by the n-heptane kinetics. The chemical mechanism, reduced to a size of 65 species and 273 elementary reactions, retains the important low/intermediate temperature ignition reactions for n-heptane, the low hydrocarbon oxidation chemistry, the formation reactions of polycyclic aromatic hydrocarbons (PAHs) (up to two aromatic rings), and the NOx formation kinetics. Numerical simulation of the transient diesel combustion process at a specific injection condition was performed. The numerical prediction shows that the current approach is capable of capturing the essential features of the diesel process such as auto-ignition and liftoff phenomena. The simulation illustrates that the lifted flame is stabilized as a triple flame. The simulated spatial soot and NO distributions are similar to those described in Dec's "conceptual diesel model". Analysis of the flame zone shows the molecular precursors of soot (e. g. PAHs and acetylene) produced during the rich burning of the sprays contributing to soot formation, whereas NO is formed closer to the oxygen diffusion layer on the lean side of the flame. The simulation was also extended to investigate the effects of charge composition variation on diesel auto-ignition and combustion. The results demonstrate that variation of oxygen molar concentration in the charge can substantially affect the auto-ignition and combustion pattern of diesel sprays.

View full abstract

Though experimental work helps in resolving many of the issues related to soot emissions, an analytical approach enables in-depth understanding of the mechanisms and principles of soot formation. In this work, an attempt is made to predict in-cylinder nitric oxide (NO) and soot concentrations using a multi-zone diesel combustion model. For estimating soot formation, the conservation equations for precursor specie, growth specie, soot volume fraction and particle number density are solved in each spray zone. The total in-cylinder NO and soot concentrations are obtained by summing up their values in each zone of the spray. The model predictions are compared with the experimental data available in literature [15]. Typical results of a comparison of energy release, cylinder pressure and in-cylinder NO and soot concentrations at a reference test condition of -8° atdc injection timing and 29.6 mg/cycle fuelling rate are shown in figures below. The predicted and experimental values of these quantities are in reasonable agreement with each other and thus confirm the ability of the proposed model to predict engine combustion and emission characteristics.

View full abstract

This study investigated the reduction mechanism of NOx in diesel combustion. Rich and high turbulence combustion was formed experimentally using a rapid compression machine with changed swirl velocity v and equivalence ratio φ, and transient concentrations of NO and lower hydrocarbons were measured at each stage of combustion by a total gas-sampling method. High-speed photography and CFD computation were also employed for the analysis of the flame behavior and NO formation. Results show that the heat release rate is proportional to the concentration of light hydrocarbons produced by the thermal cracking of fuel (Fig. A1). As is well known that NO concentration gradually increase at the initial combustion stage and, at the end of diffusion combustion, the concentration keeps maximum level. However, on the rich and high swirl condition, NO concentration decreases during the diffusion combustion (Fig. A2). Analysis of the flame behavior shows that, under the rich and high swirl condition, a ring flame is formed inside the periphery of the chamber and the flame keeps the ring structure until the end of the combustion (Fig. A3). In the ring flame region, rich and high temperature mixture is formed. A large amount of thermally cracked hydrocarbons is confined in the flame and NO formation rate decreases. These results suggest that, in the local rich and high turbulence region, NOx emission should be reduced by a chemically reduction mechanism. The mechanism is caused by some chemical species formed through the fuel decomposition.

View full abstract

Recently, many development efforts have been made to enhance the diesel spray mixing by injecting fuel at high pressure and with micro-hole nozzles in order to achieve the best compromise between the premixed and the nonpremixed stage of the combustion process. The knowledge of the instant of fuel auto-ignition under different operating conditions and the individuation of the most affecting working parameters represents, therefore, a key point for the optimisation of diesel engines design, as well as for the improvement of predictive numerical models. The auto-ignition and the first stage of combustion are some of the most complex phenomena of diesel combustion because of the occurrence of chemical reactions in a heterogeneous environment, for the characterisation of which both experimental and numerical techniques are often needed. Optical diagnostics, in fact, allows detecting and following the evolution of a limited number of chemical species representative of the auto-ignition process, whereas multidimensional numerical modelling involves, for the most, reduced kinetic scheme, in which intermediate species are not always suitable of a proper identification. In this work attention was focused on a diesel engine with an optically accessible external combustion chamber, in which a strong anti-clockwise swirl flow contributed to enhance both the spray penetration and the vapour distribution. Location and timing of auto-ignition and the first stage of diesel combustion were highlighted by flame intensity measurements from ultraviolet to visible performed in fixed positions of main interest within the combustion chamber. Chemiluminescence signals of HCO and CH radicals were detected before the indicated auto-ignition corresponding to the first combustion-induced pressure rise. OH emission was instead revealed as synchronised with auto-ignition. Experimental data relative to different air/fuel ratios were compared with results of optimised numerical simulations realised by means of a customised version of the KIVA-3 code. The concentration of the intermediate species participating the employed kinetic scheme for auto-ignition was followed with respect to time at the same spatial locations considered in the experiments. A sort of identification of these species was achieved and, at the same time, a deeper understanding of the cool flame less exothermic reactions was reached. Some insight into the transition to the hot temperature regime was derived by comparing the behaviour of both experimental and numerical data integrated over the whole combustion chamber.

View full abstract

This study is concerned with the CFD prediction of wall heat transfer in reciprocating engines, with particular reference to Diesel engines working at high peak pressures (HPP), where accurate predictions of metal component thermal and pressure loadings are required. To this end, CFD simulations are performed of flow, combustion and heat transfer in a prototype HPP Diesel engine for which detailed local time-resolved surface heat transfer measurements have been performed. Comparisons are made between these data and predictions based on two wall heat transfer treatments, one of which ignores the effects of variations of thermophysical properties within the boundary layer and another which takes them into account. The latter treatment is shown to produce substantially better agreement than the former. It is also demonstrated that conventional widely-used empirical heat transfer correlations are incapable of providing the required levels of accuracy and detail. Finally, it is shown that local heat transfer measurements also provide a stringent testing ground for spray and combustion model performance.

View full abstract

KIFP, an hexahedral unstructured version of KMB (our present CFD code for engines), has been developed. Based on KIVA's algorithms (finite volume on staggered grids, time-splitting, SIMPLE loop, sub-cycled advection...), the new solver has been built step by step with a strong control on the numerical results. This paper shows the different phases of this work. The numerical approaches and developments are discussed. Some academic examples are shown and compared with KMB or analytical results, like scalar advection or multi-species diffusion. Better precision and convergence in the physical fields are observed. Iterative loops and advective sub-cycles are also reduced thanks to the unstructured formalism. Super-scalar machines being widely used and developed, KIFP is targeted to them. OPEN-MP paradigm is used to parallelize the code, and the paper gives results on performance and speed-up (more than 3 for 4 processors in the best case). Results on a compression of a swirl motion in Direct Injection Diesel engine are given. For a given mesh density, physical results and numerical quality (driving CPU time) seems better for KIFP than for KMB. This is due to a better mesh configuration in the KIFP case (no corner cells). At last, a pancake spark ignited engine is used to compute homogeneous combustion with the extended coherent flame model (ECFM) implemented in the new solver. We find a good agreement between the results of the two codes and experiment.

View full abstract

In small bore direct injecting combustion engines a contact between liquid spray and piston or cylinder surfaces can be observed. Depending on the wall temperature the breakup and evaporating process may be enhanced or suppressed. To take this effect into account modern engine design methods apply computational fluid dynamic models that need data from experiments conducted at engine conditions. In the present work results from experimental investigations will be presented giving a large base for validating numerical models. The experimental apparatus is a high pressure injection chamber equipped with a Common-Rail diesel injection system. A wall with variable temperature aligned perpendicularly to the spray plume is used at different impingement distances. The boundary conditions are : gas temperature 600 K, wall temperature up to 600 K, gas pressure up to 50 bar and injection pressure between 500 and 1350 bar. Three NiCr-Ni surface thermocouples are soldered into the wall to spatially resolve the temperature history of the wall surface with a sampling frequency of 5 MHz. The temperature data are numerically processed to calculate the surface wall heat flux from an analytical solution of the transient energy equation. After a spray impinges on the surface it causes a fast temperature change depending on the actual droplet temperature within the spray. In the investigated cases the drops heat up from an initial liquid temperature of 340 K to 420 - 500 K prior impact depending on the boundary conditions and the wall distance. A temperature decrease of up to 10 K is detected in the stagnation region. The surface heat flux derived from these data is exceeding some MW/m^2 during the spray impact at the center. In the outer positions, no direct impingement takes place and thus the heat flux is less. The maximum flux detected at the centre increases with decreasing wall distance z_W and increasing wall temperature up to 3.0 MW/m^2 (at z_W = 15 mm), whereas at z_W = 40 mm and a wall temperature of 500 K no significant flux can be detected. The data of Phase Doppler Anemometry measurements conducted under identical conditions correlate well with the received surface heat flux measurements. At higher temperatures where the larger heat flux from the wall to the particles is observed the droplet diameter is smaller due to enhanced evaporation of the spray whereas the velocity is not influenced by the changed heat transfer conditions.

View full abstract

Turbulence plays an important role on the mixture formation and combustion processes, and consequent emission formation. In this study, the bluff body was set as a vortec generator in the combustion chamber of a DI diesel engine. A 2-dimensional unsteady computer simulation was carried out to clarify the effect of the size and shape of the bluff body and of the air stream velocity on the vortex characteristics. The computed results were compared with the experimental results obtained in a wind tunnel experimentation, where the vortex motion was visualized with smoke tracer method. In the low air velocity conditions, a recirculation zone is generated in the wake of the bluff body. Karman type of vortices then generated as the air velocity increases. In the even higher velocity region, the vortices meander and nibbling the outer air stream into the vortex region. The transition characteristics of above mentioned vortex depend strongly on the shape of the bluff body. A triangular bluff body requires higher air velocity, while the triangular prism and trapezoid prism shaped bluff bodies lower the transition velocity as shown in Fig. 1. The effects of the bluff body are then tested in a single cylinder DI engine, on the engine performance and exhaust emissions. In the engine experimentation, several different shapes of the bluff bodies were tested. For bluff body equipped engine, the smoke emission were reduced by increasing the swirl ratio, which corresponds to the increment of air stream velocity in the computer simulation and wind turned experimentation. The triangular prism bluff body produced lower smoke than the trapezoid prism bluff body equipped experimentation as shown in Fig. 2

View full abstract

Fluid motion within the cylinder of a high-speed, direct-injection diesel engine influences the entire combustion process, from the initial fuel preparation prior to ignition to the late-cycle burn-out of unburned fuel and particulates. Multi-dimensional models are increasingly being used in the design and optimization of these engines. The ability of these models to accurately predict the in-cylinder flow structures and turbulence has yet to be proven, however, particularly under fired engine operation. In this paper, a comparison of experimental and predicted flow velocities, both before and after combustion, is presented. Experimental and predicted cylinder pressure and heat release are also compared. The engine geometry employed has geometric characteristics typical of modern light-duty diesels targeted for automotive applications, i.e. : four-valves ; a central, vertical injector ; a concentric, re-entrant bowl ; and common-rail fuel injection equipment. The measurements are compared with the results of computations using STAR-CD and KIVA-3V, for the induction stroke and the compression/expansion strokes, respectively. All results reported here were obtained at an idle condition, characterized by a speed of 900 rpm and a gross IMEP of 1.2 bar. Cylinder pressure and heat-release are found to compare favorably between the experiment and the model predictions, although the predicted mass of fuel consumed in the premixed burn exceeds the measured mass burned. Similarly, the predicted in-cylinder angular momentum, obtained from a full induction stroke calculation, is found to agree to within 5% of the measured value. Just prior to the start of combustion, both modeled and experimental swirl velocity profiles are similar to solidbody fluid rotation. However, the model predicts significantly greater axial stratification of mean angular momentum than is observed experimentally. The r-z-plane mean-flow structure within the bowl is dominated by a single large-scale vortex. The location and speed of rotation of this vortex are well predicted by the model. Although RMS velocity fluctuations deep in the bowl are well-predicted at this time, larger fluctuations than the model predicts are observed experimentally near the top of the cylinder. These larger fluctuations are notably anisotropic, with the tangential component exceeding the radial component. After combustion, the axial distribution of mean angular momentum agrees more closely between model predictions and measurements. Large-scale, rotating structures in the r-z-plane, not present under motored engine operation, are also observed in both the measured and the predicted results. In contrast to the pre-combustion radial profile of tangential velocity, the post-combustion profile resembles that of a free-vortex, with the largest tangential velocities found at the inner radii. This type of profile suggests a source of turbulence production not found prior to combustion. At this time, the predicted RMS fluctuating velocities are considerably less than the measured fluctuations throughout the bowl. Measured fluctuations are again found to be anisotropic, although the radial fluctuations now exceed the tangential fluctuations.

View full abstract

This paper describes the effects of compression pressure on the spray morphology produced by a GDI injector, in terms of the cone angle and the penetration of the developing spray. Data have been collected in the form of back lit CCD images and are supported by Phase Doppler Anemometry (PDA) measurements of time averaged droplet velocity and size. The measurements were made in a high pressure bomb and two single cylinder Diesel engines with cylinder heads designed to provide good optical access to the spray. The effects of a varying compression pressure were studied by injecting into the combustion chambers at different times during the compression stroke. It was found that the cone edge could be defined from a single shot image due to the inherent stability and repeatability of this part of the spray. A family of curves have been determined, an example is shown in the bottom left figure, and a quadratic function fitted to identify the geometric boundary of the spray as a function of charge pressure. For the penetration data , however, a mean image derived from several shots was required due to the transient effects at the front of the spray giving large differences between sprays from shot to shot. The PDA data complements the images by being able to define the cone in terms of peak velocities, sample count and minimum data validation. The time of arrival for the first droplets gives an indication of the penetration of the spray. Superimposition of the velocity field onto the images provides quantitative velocity information for the different parts of the spray. A similar analysis with the drop size field provides information on the mean drop sizes present in the different parts of the spray. A combined plot of a spray image and droplet velocity vectors are shown in the figure on the right.

View full abstract

A direct injection gasoline engine employing a new stratified combustion system has been developed. The combustion strategy was achieved by a fan-shaped fuel spray and a shell-shaped piston cavity (Fig. 1). The process of stratified mixture formation and consequent stratified combustion are strongly affected by the fuel spray characteristics (Fig. 2, 3, 4, 5), therefore numerical analysis (CFD) was applied to understand the phenomena of mixture formation process. From the results of CFD (Fig. 6, 7), it was clarified that the vaporization characteristic is important in making a suitable stratified mixture formation, in addition to the spray liquid characteristics, and CFD enables prediction and analysis of actual phenomena.

View full abstract

Stratification features of DI gasoline combustion were studied, using a constant volume combustion vessel system shown in Fig. I. Indicated pressure analysis and high speed combustion observation were carried out. Though the heating efficiency η_h obtained was very low because of constant volume combustion, the relative diagnosis of initial mixture formation among design factors was judged to be possible. The factors studied were spray cone angle, swirl ratio SR and ambient temperature Ta. Ambient pressure Pa was kept constant. The effects of injection-sparking interval τ_<int> were examined. The ignition delay τ_d, defined as the time from sparking to ignition, plotted against τ_<int> were greatly scattered. This also caused the great scattering in both η_h and maximum pressure rise rate (dp/dt)_<max>. These scattering, however, were significantly reduced by plotting the data against total ignition delay τ_<tot>(=τ_<int>+τ_d), as shown in Fig. II. Clearly, both η_h and (dp/dt)_<max> decrease against τ_<tot>, and (dp/dt)_<max> decreases more rapidly than η_h. A maximum volumetric burning velocity (Sv)_<max>, which closely correlates with (dP/dt)_<max>, was proposed as the stratification degree, based on a thermodynamic consideration under such concept that higher stratification degree brings wider area of stoichiometric mixture and this result in higher burning velocity as well as less amount of unburned fuel. Fig. III compares the best values of both (Sv)_<max> and η_h, given at the shortest τ_<tot> as shown in Fig. II. for each design factor. It is noteworthy that, as spray cone angle is widened, (Sv)_<max> become higher and less affected by SR.

View full abstract

A key word to combustion control with gasoline direct injection, is "freedom of mixing", that is, no restriction on mixture preparation. One of combustion control technologies utilizing the freedom of mixing, is "two-stage combustion" for quick catalyst warm-up. It is a kind of two-stage injection, with a main injection in a late stage of the compression stroke for stratified-charge combustion, and a supplementary injection in a late stage of the expansion stroke into the burned gas, for an increase in the exhaust-gas temperature. It is an important technology for a gasoline direct injection engine to promise to meet the stringent emission regulations. Unfortunately, however, it inflicts a loss of the fuel economy, because a part of fuel is consumed not to generate the torque, but to increase the exhaust-gas temperature. It is necessary to enhance its effectiveness, and cut its operation time. A characteristic mechanism of in-cylinder combustion is "time-domain mixing". This means that in the cylinder filled with a large number of eddies, together with inhomogeneity, products in different stages of combustion process have a chance to be generated at the same time, and the products in the past, the present, and the future along by the process, is mixed up by the eddies. The successful mechanism of the two-stage combustion is controlled by the time-domain mixing. A chamber-type exhaust manifold shown in Fig. 1, is a successful technology of inducing a kind of time-domain mixing in it, and enhancing the effectiveness of the two-stage combustion. A part of exhaust gas out of a cylinder is mixed with the blow-down gas out of the following cylinder, in it. As a result of enhanced reaction in it, HC emissions are reduced, and the exhaust-gas temperature is increased, as shown in Fig. 2. The operation time of the two-stage combustion is shortened, so that the fuel economy is improved by around 0.8 % on the EURO-3 mode.

View full abstract

This paper describes the Lean Boost Direct Injection system (LBDI), a downsized gasoline engine for improved fuel economy. LBDI combines direct injection, lean operation and pressure charging, and allows a significant reduction in swept volume to be made whilst retaining vehicle performance. Engine tests have been undertaken which demonstrate the validity of the combustion system. Vehicle simulation results for typical European C class and sports utility vehicles are presented. Using the exhaust gas emission levels from engine tests and drive cycle simulation, aftertreatment requirements and configurations are considered. The alternative of an advanced small bore HSDI is compared.

View full abstract

On two engines, that are shown in the title, SI-Stratified-GDI specify MITSUBISHI-SI-GDI and CI-Pre-Mixed-DI is NISSAN-CI-DI-MK Engines. The discussion in this paper is limited in the low and middle load zone without catalyst, Searched points to have not been clarified, would be written as follows ; In GDI, 1) The radical forming in GDI' spray flame[M-1〜3] suggested this would be generated under HGF condition. 2) To fulfill 1), the concept of the CVCC-REABP phenomenon[Ma-1, 2] was adopted. 3) To fulfill for the REABP phenomenon of 2), STIM that is the earlier burned layer fed back to the unburned layer has been presumed. 4) High-speed rotating stratified vortex in MITSUBISHI-GDI fulfilled the demand of 3). Next in MK, 5) Stratified swirl of Swirl Ratio=12 and extraordinary retarded ignition delay suggested the formation of STIM & REABP phenomenon shown in 2, 3), also in the same way the formation of HGF condition and radical formation under such similar physical and chemical condition, as shown above in GDI were expected. 6) These concepts of STIM, REABP and radical forming were the key of making the causality to clear this SRFNP in these two engines. The pur-pose of this paper is to predict by lighting up onto both combustion mechanisms from a quite different view angle as ever. Then, it will anticipate some contribution still quantitative but it clarifies the causality from the foreseeing situation on the development of this kind of stratified combustion engine. Enumerated conclusion is shown as follows ; 1) MK was also stratified similar as GDI with high-speed swirl. 2) In GDI, the duration after the injection finished to the ignition timing which has created MK zone acted the important factor as same as MK. 3) The complicated STIM, either not propagation nor diffusion combustion much affected to SRFNP. 4) In GDI and MK accompanied by STIM and REABP, then the states LGF and HGF states were formed. 5) Measured radical in GDI was born through HGF with REABP phenomenon as shown in 4) and the forming of radical in MK would be similarly done as in GDI. 6) In GDI and MK, there was a combination of external EGR, the rotating stratification flow, the ignition timing, STIM and REABP phenomenon commonly. With these, it became the forming of HGF and radical. These contributed on the attainment of SRFNPH and this mechanism in both combustion engines, that the author foreseen would be in the high grade of possibility in things above mentioned.

View full abstract

It is well known that in a four-valve spark ignition engine with pentroof combustion chamber, the tumbling flow is broken down into small eddies in the latter stage of the compression stroke under the condition of higher compression ratio and as the turbulence intensity increases. However, many characteristics of the turbulence at the time when the tumbling flow is broken down are not yet understood. Generally, the fractal dimension of the turbulence is effective for showing the eddy structure. However, when the turbulence is analyzed with this ordinary method, only a statistically averaged value is obtained, and it is difficult to evaluate the change of the eddy structure when the tumbling flow breaks down. In this study, we propose a new method to calculate the fractal dimension by using a wavelet transform which is a time-frequency analysis method and try to evaluate the change of the eddy structure in the turbulence when the tumbling flow is broken down. The procedure of our method is 1) Carry out the wavelet transform of the turbulence and obtain the information of the turbulence about both time and frequency. 2) Calculate the fractal dimension at each time from the wavelet transform of the turbulence. 3) Evaluate the change of the eddy structure in the turbulence by using the fractal dimension obtained. The Figure shows the result in the case of compression ratio ε=5.5, where Fig. (a) is the wavelet transform of the turbulence and Fig. (b) is the result of the fractal analysis. As shown in Fig. (b), the fractal dimension decreases a little near 320 degrees crank angles because the tumbling flow is broken down and the turbulence velocity becomes large as shown in Fig. (a). Then, the fractal dimension increases and becomes D_ω=1.591 near TDC because the power of the small eddies become larger. That is, the eddies that were generated by the tumbe flow broken down have larger scale and transmit the energy which they have to the small eddies in the compression stroke. After the TDC, the gas flow is also generated by the piston motion and the fractal dimension increases as shown in Fig. (b). Therefore, based on the above discussion, the fractal dimension D_ω proposed is effective to evaluate the change in structure of the eddies quantitatively.

View full abstract

Direct injection spark ignition (DISI) engines, which are able to achieve better thermal efficiency and higher output power simultaneously, have been developed. The DISI engine is operated unthrottled in an ultra-lean condition by distinctively stratifying the charge and by preparing a fuel-rich mixture around the spark plug. However, the effects of mixture inhomogeneity and simultaneous details of the flow field at the time of the spark on the combustion have not yet been clarified. In this research, the fundamental flame propagation and combustion characteristics of a stratified fuel-air mixture, similar to mixtures used in direct injection spark ignition (DISI) engines, were investigated. A swirling fuel-air mixture was produced in a disc-type combustion chamber by accelerating the mixture tangentially. Propane was injected into the center of the combustion chamber at an injection pressure of 0.3 MPa. The strong swirl combined with the gas injection created an axisymmetric fuel-rich mixture near the spark location and a lean mixture near the wall. The swirl, gas injection and spark timing changed the flow field, turbulence intensity, and inhomogeneity of the fuel-air mixture. The mixture was ignited at the center of the vessel. The fuel concentration distribution and the swirling flow field at the time of the spark were measured using LIF and LDV, respectively. Experimental tests were performed for various fuel distributions and different swirl flow at the ignition timing. Flame propagation and the combustion characteristics of an inhomogeneous fuel-air mixture are discussed using instantaneous flame images obtained from a high-speed video and the pressure history. Three main conclusions are drawn from this work. First, strongly swirling flow with gas injection at the center generates successive inhomogeneous fuel concentrations. By changing the period between gas injection and ignition, various gradients of inhomogeneous mixture concentrations and equivalence ratios can be obtained in the vicinity of the spark location. A stratified charge with an appropriate mixture near the spark location allows the flame to propagate and combustion to occur, even when an ultra-lean fuel-air mixture is used. Second, the stratified stoichiometric mixture in the vicinity of the spark location propagates more rapidly than in the homogeneous case with the same overall equivalence ratio, resulting in a higher rate of pressure increase and a shorter ignition delay time. Third, when stratified mixtures with the same pattern of fuel concentration but different turbulence intensities are considered, the main combustion period decreases with turbulence intensity at an overall equivalence ratio of 0.30, since the burning velocity also increases.

View full abstract

The stratified charge combustion system has been widely studied due to the significant potentials for low fuel consumption and low exhaust gas emissions. The mixture formation process for a direct-injection stratified charge engine is influenced by various parameters such as the atomization, the fuel evaporation, and the in-cylinder gas motion at high temperature and high pressure conditions. It is difficult to observe the in-cylinder phenomena in such conditions and also challenging to analyze the following stratified charge combustion. Therefore, the combustion phenomena in idealized stratified charge conditions aiming to analyze the fundamental stratified charge combustion are examined. That is, an experimental apparatus which can control the mixture distribution and the gas motion at ignition timing was developed, and the effects of turbulence intensity, mixture concentration distribution, and mixture composition on stratified charge combustion were examined. The schematic of experimental setup is shown in Fig. 1. Three volumes separated by two partitions were evacuated and then three premixtures were charged simultaneously through pressure regulators. After the charging was completed, the gas motion generator was operated to control turbulence at spark timing. After partitions were removed, ignition was made. As a result, it was found that in stratified charge combustion using propane-air mixture, when the overall equivalence ratio, φo, is unity, no advantages by charge stratification is found, but when φo is set at 0.8 combustion is enhanced by additive effects combining turbulence with charge stratification, and that combustion is enhanced by strong turbulence in slightly stratified charge fields using methane-air mixture, but extending stratification brings unfavorable results compared with the homogeneous condition.

View full abstract

During the last decade, a new approach to multi-dimensional computations of premixed turbulent combustion has been developed. It consists of the use of an analytical expression for the local turbulent flame speed in order to close the averaged balance equations describing the combustion process. Several models have been suggested utilizing this approach and a number of successful applications to 3D computations of combustion in S. I. engines was reported at COMODIA 98. Although a submodel of turbulent flame speed, S_t, is at the core of this approach, a consensus has not been reached on this issue. The choice of the most reliable submodel is strongly impeded by the wide scatter of the published experimental data on S_t. Moreover, this scatter questions the physical correctness and practical usefulness of the concept of turbulent flame speed. The main goal of this work is to support the concept by discussing certain physical sources of the scatter and by outlining methods of reducing it. Three physical mechanisms are discussed in the paper. First, numerous experimental data indicate that typical premixed turbulent flames are developing flames, rather than fully developed ones, i. e., the mean turbulent flame brush thickness, δ_t, is not constant but grows permanently either with time in expanding flames or with the distance from the flame-holder in stationary flames. Due to the growth of δ_t, the speeds associated with different reference surfaces inside the flame brush can be substantially different, e. g., the difference can be as large as u'. A new method of unambigouosly defining the reference surface, the speed of which straightforwardly characterizes the mass burning rate, is developed by theoretically analyzing self-similar (i. e., spatial profiles of the mean progress variable, measured at different flame development times and presented in the dimensionless form by using δ_t(t), are collapsed to a universal curve) premixed turbulent flames in the statistically planar, one-dimensional case. This reference surface is characterized by the following reference value of the Favre-averaged progress variable [numerical formula] ; [numerical formula] ; [numerical formula], where ρ(ε)≡ρ^^-((x-x_f)/δ_t)/p_u is a universal dimensionless profile of the mean density (ρ(-∞)=1) and x_f(t) s the flame position. Numerical simulations indicate that the speed of this reference surface is equal to the mass burning rate divided by p_u and, thus, support the proposed method. Second, the mass burning rate in expanding spherical flames is well known to develop as the flame kernel grows, because, in particular, it experiences a wider range of the turbulence spectrum. Simple estimates discussed in the paper show that such transient effects can change the dependence of flame speed on turbulence characteristics not only quantitatively but even qualitatively. To suppress these effects, flame development time should be sufficiently large (e. g., larger than the turbulence time scale). Third, the speed of a statistically spherical flame differs from the speed of the planar one, other things being equal, due to (1) the mean flame curvature effects, and (2) the reduction in the maximum gas flow velocity induced by hot combustion product expansion, due to a wide flame thickness. In laminar flames, such effects are well known but weak. In turbulent flames, the effects can be substantial, because δ_t, is much larger than the laminar flame thickness. By analogy with the laminar combustion, the following equation [numerical formula], is proposed to be used in order to evaluate a fully developed turbulent flame speed, S^o_t, by processing the dependencies of the mean flame radius, r_f, on time, t, measured in expanding spherical flames. Here, Ma_t is a turbulent analog of the Markstein number used widely in laminar flame studies. This processing has been applied to the published experimental data. The

(View PDF for the rest of the abstract.)

View full abstract

It is well known that the combustion in SI engines typically takes place in the wrinkled laminar flame region. According to the concept of the conventional models of turbulent burning velocity in this turbulence condition, the ratio of the turbulent burning velocity to the local burning velocity is proportional to the ratio of the turbulent flame surface area to the laminar flame surface area. The area ratio is well approximated to be proportional to the ratio of the turbulence intensity to the local burning velocity, which is usually taken to be equal to the original laminar burning velocity. However, in our previous works, the mean local burning velocity turned out to be changed from the original laminar burning velocity due to the preferential diffusion effect and it was found to be an important factor as dominating the turbulent burning velocity. This fact seems to contradict the concept of the conventional models. In the present study, an attempt is made to investigate directly the local propagation characteristics of methane and propane premixed turbulent flames in the wrinkled laminar flame region. A laser tomography technique is used to obtain the local flame configuration and movement. And quantitative analyses are performed. The local flame front curvature 1/r and the local flame burning velocity S_F are quantitatively obtained as the key parameters of the turbulent combustion. In the first place, the observation of sequential flame tomograms shows that the turbulent flame front can be classified into the active and inactive part depending on its geometric configuration. In the next place, the values of S_F are obtained directly from the sequential flame tomograms as the local burning velocity. Figure shows a typical probability density function (pdf) of the acquired 1/r normalized by the preheat zone thickness η_o, and a distribution of the acquired S_F normalized by S_<LO> for the laminar flame and the turbulent flame, respectively. It is found that the values of S_F of the turbulent flame are not a constant value as S_<LO>, but to be distributed over a wide range depending on the local curvature of the turbulent flame. In addition, the mean value of S_F for the methane mixture has a tendency to become larger as decreasing its equivalence ratio, whereas that of the propane mixture to become smaller.

View full abstract

Natural gas is now considered to be one of the most promising alternative fuels for automotive engines because of the low exhaust emission characteristics and very large deposits spread around the world, much of which is not exploited. So it is important to clarify the nature of this energy source. The purpose of this paper is to examine the turbulent combustion properties of methane, which is the main component of natural gas, considering the application to spark-ignited engines. The below left figure shows the experimental result on turbulent burning velocities, S_T, as a function of turbulence intensity, u', where both values are non-dimensionalized by the corresponding laminar burning velocities, S^o_u, using a constant-volume combustion bomb. From the figure, it is found that for methane mixtures the leaner mixtures become, the larger the turbulent burning velocities even at the same turbulent intensity. We have inferred that this effect is caused by the change in local burning velocity due to the flame stretch to which each local flamelet in a turbulent flame is subjected. In this case, the local burning velocity of each flamelet changes from the unstretched laminar burning velocity with increasing stretch near-linearly. According to the asymptotic analysis, general relation can be expressed as the below equation. [numerical formula] Where S_u is stretched laminar burning velocity, Ka is non-dimensional flame stretch, characterized by Karlovitz number. So it is found that Ma, which is referred to as Markstein number, represents the sensitivity of local burning velocity to flame stretch. At the same value of Ka, mixture with the smaller value of Ma has the larger value of S_u/S^o_u. So in turn we have investigated the correlation between Markstein numbers and turbulent burning velocities. The below right figure shows the variations of Markstein numbers with equivalence ratio, which were obtained in this study and cited from other studies. From the figure, for methane mixtures Ma is found to decrease with decreasing equivalence ratio, indicating that the stretched burning velocity is estimated to relatively increase on fuel leaner region. These trends quantitatively agree with those of turbulent burning velocities, as seen in left figure. From the above discussion, it can be concluded that methane intrinsically favors lean burn combustion. And this can be explained by the difference in sensitivity to flame stretch.

View full abstract

Laminar flame propagation in a stratified charge with a step function distribution of equivalence ratio was studied in a spherical combustion vessel. The stratification was created by trapping a methane air mixture at fuel equivalence ratio φ_1 in a small spherical soap bubble at the center of the vessel ; the remaining of the vessel was filled with the mixture at fuel equivalence ratio φ_2. The charge was ignited at the vessel center by a focused laser beam. Results were obtained at various combinations of φ_1 and φ_2. The results show that the flame has a memory of the previous history - it takes a substantial amount of time for the flame speed to relax from its steady state value at φ_1 to that at φ_2. A 1-D, time-dependent numerical simulation of the flame propagation in a methane/air charge with step-stratification was used to explain the observations. For a rich-to-lean flame transition, the gradient transport of heat and radicals from the previously burned gas tend to give a "back-support" to enhance the flame speed over the nominal lean value for a homogeneous charge. Thus the usual notion of laminar flame speed being determined by the unburned mixture condition does not apply to a stratified charge in many practical situations such as in a stratified charge engine.

View full abstract

Combustion characteristics of the transient gas fuel jet were investigated in this study. Stratified mixture was formed in the constant volume bomb by the injection of gaseous propane into the quiescent air in the bomb. The amount of the injected fuel was set to 0.07 in overall equivalence ratio. It was ignited by the electric spark. The ignition timing had large influences on the combustion pressure and the amount of the unburned fuel. The amount of the unburned fuel in the burned gas was larger as the ignition timing was later regardless of the ignition positions. Flame did not propagate to the lean region in the cases of late ignition timings. Fuel in this lean region remained unburned. Bulk quenching in the over lean region seems to be the main cause of the unburned fuel because the over lean region will increase as ignition timing is late. The flame propagation in the case of the ignition during the fuel injection was quite different from those in the cases of the ignition after the end of the fuel injection. Secondary flame came out in the burned region after the initial flame had propagated through the fuel jet only in this condition. Its shape was quite different from the initial one in the early stage of the combustion process. And it was luminous and showed red. This secondary flame remained there over 80ms. It seemed that the rich region surrounded by the burned gas and the fuel injected after ignition burned gradually mixing with air entrained into this region after the initial flame propagated throughout the fuel jet. The limit of flame propagation to the lean region was investigated next as the main cause of the unburned fuel was bulk quenching in the over lean region. In order to examine at what fuel concentration flame quenching occurred, lean fuel-air mixture below the flammability limit was supplied as the charge instead of air. Fuel was injected into this mixture and ignited. Though the mixture charge itself was too lean to ignite by the spark ignition, flame initiated in the fuel jet propagated to the large area compared to the case of no fuel addition to the charge. More propane than the amount of the main injection was burned when the equivalence ratio of the charge before fuel injection, φ_c was 0.40 or more. The amount of the burned fuel normalized by the amount of the main fuel injection, N_B increased considerably in the range of φ_c=0.40 to 0.54. The limit of flame propagation to the lean region seemed to be around this range.

View full abstract

Fuel mixture distribution has been measured during the mixture formation and the combustion period by the LIF technique in order to analyze the combustion fluctuation in a DI gasoline engine. Fig. A1 shows the cycle-by-cycle variation of the equivalence ratio at the spark position just before the spark timing and its effect on the initial combustion period and the IMEP value. From the figure, it can be seen that the equivalence ratio is highly fluctuated even in the same injection timing. At the advanced (-70°) or the retardant (-58°) injection timing, the initial combustion period tends to long and the IMEP value becomes low when the leaner cycles (φ<1) or the richer cycles (φ>2) appear. From these results, it has been revealed that the combustion fluctuation at both the advanced and retardant injection timing is dominated by the mixture concentration at the spark position and timing. In contrast to this, when the injection timing is set at the best combustion fluctuation condition (-63°), the initial combustion period is not affected by the fluctuation of the equivalence ratio. In this condition, the cycle-by-cycle IMEP value has a correlation not with the initial combustion period but with the main combustion period. Therefore, the combustion fluctuation is dominated by the latter stage of the combustion. Fig. A2 shows the fuel distribution during the combustion period. From the LIF image, the latter stage of the combustion occurs at the edge region of the piston cavity. Fig. A3 shows the cycle-by-cycle LIF intensity, which is mainly proportional to the fuel quantity, and its effect on the IMEP value. From this figure, it is evident when the unburned fuel remains much, the IMEP value is low ; strong correlation can be seen in it. Therefore, it can be said that the combustion fluctuation near the best injection timing is due to the cycle-by-cycle variation of the unburned fuel existing at the cavity edges during the latter combustion period. Based on this analysis, it became possible to improve the combustion fluctuation by the procedure for the reduction of the over lean mixture at the cavity edge region.

View full abstract

The optimization of the vaporization and mixture formation process is of great importance for the development of modern gasoline direct injection (GDI) engines, because it influences the subsequent processes of the ignition, combustion and pollutant formation significantly. In consequence, the subject of this work was the development of a measurement technique based on the laser induced fluorescence (LIF), which allows the two dimensional visualization and quantification of the in-cylinder air/fuel ratio. The investigations were carried out in a transparent engine provided by the BMW Group, Munich. A substitute fuel consisting of benzene and triethylamine dissolved in the non-fluorescent base fuel isooctane has been used. The main property is that the fluorescence signal is proportional to the equivalence ratio (φ=1/λ) at low absorption and negligible influences of the residual gas components CO_2 and H_2O at engine like conditions. The calibration of the LIF-signal was performed directly inside the engine, at a well known mixture composition, immediately before the direct injection measurements were started. Thus, it was possible to minimize crucial effects caused by window fouling, the non-uniformity of the laser illumination and finally the optical particularities of the setup. For comparison with the LIF results, Raman measurements were carried out simultaneously. The Raman scattering was selected as a reference measuring method, since the results of this technique are within the scope of our experiments nearly independent on ambient conditions as pressure and temperature, as well as on window fouling. In addition to different engine operating points in the homogeneous regime, the mixture formation was also investigated at stratified charge near the ignition point. A comparison between the results of both measurement techniques shows very good agreements of the air/fuel ratios.

View full abstract

The unburned end-gas temperatures in a combustion chamber of a conventional 4-cylinder DOHC spark-ignition engine were measured using the broadband CARS temperature measurement technique. The test engine was fueled with primary reference fuel 80 and fuels with research octane numbers of 70.9, 83.4, 91.5 and 100.4. The measured CARS temperatures were compared with adiabatic core temperatures calculated from measured pressures. Significant heating by pre-flame reaction in the end gas zone was observed in the late part of compression stroke under both knocking and non-knocking conditions. CARS temperatures measured at the cylinder pressure of 1400 kPa and above were higher than adiabatic core temperatures. These results indicate that some exothermic reactions exist in low pressure and temperature regions. The CARS temperatures began to be higher than the adiabatic core temperature when the end-gas temperatures reached 700 K. The temperature elevation due to the pre-flame reaction correlated better with the CARS temperature than with the cylinder pressure. The temperature elevation showed very good correlation with the unburned gas temperature without regard to the research octane number of fuel.

View full abstract

The present work gives a further contribution to the heat release analysis of the measured cylinder pressure time-history, that is the most commonly used combustion diagnostic for determining the actual burning rate in spark-ignition engines. A critical examination was made of the thermodynamic models applied for such an analysis, paying specific attention to their calibration techniques, to the simplifications adopted and to the correlations used for evaluating the bulk gas-wall heat transfer. With respect to these latter, the convective surface-averaged heat flux is usually determined by a quasi-steady application of Newton's law, regarding the heat transfer coefficient as a function of the instantaneous flow properties as well as of geometric and operating engine variables. In a previous paper, the authors proposed and applied a more general complex-variable formulation of Newton's law of convection for modeling the instantaneous surface-averaged heat flux so as to take the unsteadiness effect of the gas-wall temperature difference into account. A refined two-zone heat release model for cylinder pressure-data reduction was developed and assessed. As in the former version of the model, the thermodynamic properties of both reactants and products were evaluated from JANAF tables with a multiple species equilibrium composition calculation performed for the burned zone. The novelties of the present model include the following improvements and their combined implementation. In addition to the new unsteady convection model, a CAD procedure was introduced in order to estimate the burned- and unburned-zone heat-transfer wall areas for assigned geometric features of the flame front, whereas the burned- and unburned-gas volumes are often used as weighting factors of their related heat fluxes in the global heat-transfer calculation. Furthermore, the energy conservation equation was applied to the unburned-gas zone instead of the isentropic relation that is commonly used for evaluating the temperature of the unbumed gas. A refinement was also included for the calibration procedure of the cumulative mass-fraction burned, or of the fuel-energy released, at the end of the flame propagation process. Usually, the predicted mass-fraction burned after combustion is completed is matched to the measured combustion efficiency by adjusting the heat-transfer correlation constants. More specifically, the amount by which the theoretical maximum mass-fraction burned is taken to be less than unity is set equal the unreleased energy fraction determined from either the measured HC or the measured HC, CO and H_2 in the engine exhaust. This lost energy is evaluated as a fraction of the total fuel energy available. In the proposed heat-release model, the heat-transfer calibration is made through an overall energy balance of the whole cylinder charge during the combustion process. The unreleased energy predicted at the end of the flame propagation is correlated to the combustion efficiency determined from the exhaust gas composition. The new heat-release model was applied to the analysis of the combustion parameters in a multivalve SI engine fueled by either gasoline or CNG under a significant sample of operating conditions. The model appeared to be an accurate tool of combustion diagnostic for SI engines, being also relevant to diesel engine application.

View full abstract

The combination of physical models including a combustion model of an advanced engine control system (Fig. A-1) was proposed to obtain sophisticated combustion control in lean mixture combustion and high boost engines, including homogeneous charge compression-ignition and activated radical combustion with variable intake valve timing and a supercharger. Physical intake, engine thermodynamic, and combustion models predicted mass flow rate and exhaust gas recycle rate in the intake system, and temperature and pressure in the cylinder, based on the signals of an air flow sensor and a pressure sensor. Then, These models determined control variables such as air mass, fuel mass, exhaust gas recycle valve opening, intake valve timing and combustion start crank angle, resulting in combustion improvement in the above conditions. The combustion characteristics and auto-ignition parameters in the combustion model were investigated, and compared with some experimental data. Fig. A-2 shows air mass G_a, fuel mass G_f, and combustion start crank angle θ_i versus indicated mean effective pressure P_<mi> to attain optimum combustion. When p_<mi> was low, G_a/G_f was 20 and the mixture was lean, When the p_<mi> was high, G_a/G_f=15, mixture was stoichiometric. When p_<mi> became high enough, the stroke volume was decreased by using late intake valve closing, and knock tendency decreased. Therefore, p_<mi> increased by ignition timing advance (decrease in θ_i). Total calculation times per intake stroke depended on the time step of the intake model and the crank angle step of the combustion model. A calculation time within 5ms was attained.

View full abstract

The development and optimization of internal combustion engines requires the application of advanced development tools. In addition to experimental methods, numerical calculations are needed in order to obtain an insight into the complex in-cylinder processes. In this context, the modeling of the combustion phenomena represents an important aspect. Therefore the objective of this paper is to present numerical methods to analyse the combustion process in premixed spark-ignition engines. The investigations were performed in a 6-cylinder 2.8 1 SI-engine running at wide open throttle. The numerical calculations were performed using the finite volume CFD code STAR-CD. The mesh generation process, including the description of the piston and the valve motion, was automated using ICE. Combustion in the present study was treated with the one-equation Weller flamelet model. This model was implemented in Star-CD. The mass fractions of the combustion products were assumed to follow the local and instantaneous thermodynamic equilibrium values. The equilibrium composition of the cylinder charge was calculated according to Olikara/Borman. Eleven species were considered : O_2, CO_2, H_2O, N_2, H, O, N, H_2, OH, CO and NO. Isooctane was used as fuel. For the calculation of the convective heat transfer during the combustion process a further submodel for the calculation of the heat transfer coefficient was used. In this work, different operating conditions were analysed. For all operating conditions the gas exchange process and the combustion process were calculated. Every calculation started 40° CA BTDC and finished when the combustion was completed. The boundary conditions were gained by experimental investigations. For the verification of the combustion model, calculated cylinder pressure data and mass fractions burned are compared to experimental results. The results of the combustion process are discussed for different engine speeds and equivalence ratios. This discussion reveals that the combustion model used shows encouraging results. The comparison of the calculated and measured in-cylinder pressure indicates good agreement for equivalence ratios between 0.87 and 1.25 and engine speeds up to 3000 r/min. The shape of the predicted flame appears to be reasonable.

View full abstract

Owing to continuing air pollution problems, stringent regulations are being enforced to reduce unburned hydrocarbon (HC) emissions from spark ignition engines. A number of attempts have been reported on the sources of HC emissions. The crevices are a major source of unburned HC emissions in spark ignition engines. The largest of crevice regions is the piston-ring crevice, however head-gasket, spark plug, and valve seat crevices are not insignificant. After the end of primary engine combustion, some of unburned HCs from crevices are oxidized upon mixing with hot burned gases during the expansion and exhaust processes. The others are emitted to the exhaust port or retained in the cylinder with the residual gases to be recycled in the next cycle. Some of unburned HCs that leave the cylinder are oxidized at the exhaust port. The oxidation process is considered as the dominant HC removal process during engine start-up and warm-up periods. Therefore, quantifying the HC oxidation in the cylinder is potentially important, but a systematic investigation of the HC oxidation has not been conducted due to the difficulties and limitations of engine experiments. A 3-dimensional simulation was developed to predict the oxidation rate of unburned HCs in combustion chamber of a propane-fueled spark ignition engine with consideration of flow, mixing, and heat transfer. The computational moving mesh with the piston and head-gasket crevices was constructed for a commercial 4-valve spark ignition engine. A FAE premixed turbulent combustion model and a flame wall quenching model were applied to simulate flame propagation. In order to predict the unburned HC oxidation, a 4-step oxidation model was used. The mixing, oxidation, and exhausting of unburned HCs are examined, and the relative importance of different crevices was studied. 68.1% of fuel from the crevices are oxidized during the expansion and exhaust processes, and ethylene corresponding 17.8% of oxidized fuel are produced by the fuel oxidation. Head-gasket crevice HCs are exhausted early from blowdown period, on the other side, highly concentrated piston crevice HCs are mainly emitted in the end of the exhaust process. Different locations of crevices influence the oxidation rates, exhaust timings and exhaust degrees. The THC oxidation rate of the piston and the head-gasket crevices are 61.3% and 68.8% respectively. Consequently, the piston crevice contributes to 82.4% and head-gasket crevice does to 17.6% of the engine-out THC emissions relatively. The relative importance of different crevices must be evaluated by considering their shapes, positions, movements, exhaust processes, and engine operating condition as well as volumes, even though crevice size is considered as the most important respect.

View full abstract

The interaction between thermo-fluid dynamics and auto-ignition reaction was analyzed by the CFD simulation integrating the low temperature oxidation reaction. In the first analysis, a compression auto-ignition of a premixed mixture in a rapid compression machine was solved in a laminar axisymmetric field. The computed results show that the temperature field becomes non-uniform at the end of compression, being higher in the outer region and lower in the central region ana near wan regions, This is due to the vortex induced in the piston-liner corner and wall heat transfer. It is shown that the auto-ignition initiates different places depending on whether local temperatures are outside or inside the negative temperature coefficient regime of the low temperature oxidation reaction. In the second, the auto-ignition in an s. i. engine was analyzed in a three dimensional turbulent field by solving the spatially filtered transport equations instead of the Reynolds averaged ones in consideration of the strong non-linearity to temperature in the reaction terms in the fluctuating temperature field. The computed results for a disk-shaped combustion chamber with a central spark plug show that temperature fluctuations appear during compression due to random variations in the wall heat transfer which were generated by the random motion flow, though a uniform temperature field was assumed as the initial condition. Figure 1 shows the computed pressure compared with the experimental results. The calculation terminated at the auto-ignition initiation, which was assumed as the time when any local temperature of the unburned mixture reached 1100 K. The predicted time of the first local auto-ignition appearance is seen to be 1.1 deg. BTDC, which falls in the range of the cycle variation in experimental results. The computed temperature fields of the unburned mixture at this time are shown in Figure 2. The figure demonstrates that the auto-ignition sites appear sporadically as seen in the high-speed photographs presented in the literature.

View full abstract

A multi-dimensional combustion model combined with a detailed kinetics has been developed by the link between KIVA-3 and CHEMKIN-II with some modifications. In this model, various elementary reactions can be taken into account in a common CFD code. On the other hand, Homogeneous Charge Compression Ignition (HCCI) combustion has been attracting growing attention in recent years due to its potentials for simultaneously improving exhaust gas emissions and fuel consumption in both gasoline and diesel engines. However, it is difficult to control occurrences of violent pressure increases at the time of ignition in the HCCI combustion. In this paper, a numerical study was carried out to investigate the chemical reaction phenomena encountered in the HCCI combustion process of natural gas by the use of a developed CFD detailed kinetics model. In this model, 151 chemical species and 501 elementary reactions are taken into account. Fig. 1 shows that pressure, RHR and major chemical species mass histories in the cylinder as a whole obtained by a multi-dimensional model. As shown in this figure, hydrocarbons having higher carbon numbers are consumed ahead of time. During low-temperature reactions, HCHO and CO are formed gradually. When the high-temperature reaction occurs to accompany rapid pressure increases, these chemical species disappear. On the other hand, OH and CO_2 are formed rapidly at this term under this condition. Moreover, other results indicate iso-contours of temperature and various chemical species in the cylinder and the effects of its heterogeneity on HCCI ignition.

View full abstract

To investigate the process of spark ignition in a methane-air mixture, numerical analysis with elementary reactions and ion-molecule reactions has been performed. For this investigation, two chemical reaction schemes are employed : One, defined as Scheme A, contains 32 reaction steps. The other, defined as Scheme B, contains 81 reaction steps. It is concluded that minimum ignition energy of Scheme B is a little higher than that of Scheme A, because the formation of ion-molecules needs much heat absorption. Comparing with experimental data, it is said that Scheme B is better simulation of real spark ignition. It is also concluded that a pathway for the production of CHO^+ and H3O^+ during discharge period is important for spark ignition.

View full abstract