Search Results

The challenge of developing a robust, real-time driver gaze classification system is that it has to handle difficult edge cases that arise in real-world driving conditions: extreme lighting variations, eyeglass reflections, sunglasses and other occlusions. We propose a single-camera end-toend framework for classifying driver gaze into a discrete set of regions. This framework includes data collection, semi-automated annotation, offline classifier training, and an online real-time image processing pipeline that classifies the gaze region of the driver. We evaluate an implementation of each component on various subsets of a large onroad dataset. The key insight of our work is that robust driver gaze classification in real-world conditions is best approached by leveraging the power of supervised learning to generalize over the edge cases present in large annotated on-road datasets.

This paper evaluates the lag time in a turbocharged single cylinder engine in order to determine its viability in transient applications. The overall goal of this research is to increase the power output, reduce the fuel economy, and improve emissions of single cylinder engines through turbocharging. Due to the timing mismatch between the exhaust stroke, when the turbocharger is powered, and the intake stroke, when the engine intakes air, turbocharging is not conventionally used in commercial single cylinder engines. Our previous work has shown that it is possible to turbocharge a four stroke, single cylinder, internal combustion engine using an air capacitor, a large volume intake manifold in between the turbocharger compressor and engine intake. The air capacitor stores compressed air from the turbocharger during the exhaust stroke and delivers it during the intake stroke.

The combustion process after auto-ignition is investigated. Depending on the non-uniformity of the end gas, auto-ignition could initiate a flame, produce pressure waves that excite the engine structure (acoustic knock), or result in detonation (normal or developing). For the “acoustic knock” mode, a knock intensity (KI) is defined as the pressure oscillation amplitude. The KI values over different cycles under a fixed operating condition are observed to have a log-normal distribution. When the operating condition is changed (over different values of λ, EGR, and spark timing), the mean (μ) of log (KI/GIMEP) decreases linearly with the correlation-based ignition delay calculated using the knock-point end gas condition of the mean cycle. The standard deviation σ of log(KI/GIMEP) is approximately a constant, at 0.63. The values of μ and σ thus allow a statistical description of knock from the deterministic calculation of the ignition delay using the mean cycle properties

The fuel effects on HCCI operation in a spark assisted direct injection gasoline engine are assessed. The low load limit has been extended with a pilot fuel injection during the negative valve overlap (NVO) period. The fuel matrix consists of hydrocarbon fuels and various ethanol blends and a butanol blend, plus fuels with added ignition improvers. The hydrocarbon fuels and the butanol blend do not significantly alter the high or the low limits of operation. The HCCI operation appears to be controlled more by the thermal environment than by the fuel properties. For E85, the engine behavior depends on the extent that the heat release from the pilot injected fuel in the NVO period compensates for the evaporative cooling of the fuel.

The performance of an extended stroke spark ignition engine has been assessed by cycle simulation. The base engine is a modern turbo-charged 4-stroke passenger car spark-ignition engine with 10:1 compression ratio. A complex crank mechanism is used so that the intake stroke remains the same while the expansion-to-intake stroke ratio (SR) is varied by changing the crank geometry. The study is limited to the thermodynamic aspect of the extended stroke; the changes in friction, combustion characteristic, and other factors are not included. When the combustion is not knock limited, an efficiency gain of more than 10 percent is obtained for SR = 1.5. At low load, however, there is an efficiency lost due to over-expansion. At the same NIMEP, the extended stroke renders the engine more resistant to knock. At SR of 1.8, the engine is free from knock up to 14 bar NIMEP at 2000 rpm. Under knocking condition, the required spark retard to prevent knocking is less with the extended stroke.

Comparisons have been made between dual-fuel (80% port-injection gasoline and 20% direct-injection diesel by mass) Highly Premixed Charge Combustion (HPCC) and blended-fuel (80% gasoline and 20% diesel) Low Temperature Combustion (LTC) modes on a 1-L single-cylinder test engine. In the HPCC mode, both early-injection (E-HPCC) and late-injection (L-HPCC) of diesel have been used. The comparisons have been conducted with a fixed fuel injection rate of 50 mg/cycle at 1500 rpm, and with the combustion phasing fixed (by adjusting the injection timing) so that the 50% heat release point (CA50) is at 8° ATDC. The rapid heat release process of LTC leads to the highest maximum pressure rise rate (MPRR). A two-peak heat release process is observed in L-HPCC, resulting in a lower MPRR. The heat release rate and MPRR values for the E-HPCC are comparable to the L-HPCC values. The EHPCC mode provides the lowest NOX emission. The soot emissions for all three modes are low.

A twin-land oil control ring (TLOCR) model is used to evaluate TLOCR friction and the results are compared to the experiment measurement in a single cylinder floating liner engine under motoring condition. The model is based on a correlation between the hydrodynamic pressure and film thickness, which is generated using a deterministic model. The well-known three-regime lubrication is predicted with the model for ring with different ring tensions under various engine running conditions. A good match is found for the model and experiment results.

Initial measurements of water vapor temperature using a Fourier domain mode locking (FDML) laser were performed in a carefully controlled homogenous charge compression ignition engine with optical access. The gas temperature was inferred from water absorption spectra that were measured each 0.25 crank angle degrees (CAD) over a range of 150 CAD. Accuracy was tested in a well controlled shock tube experiment. This paper will validate the potential of this FDML laser in combustion applications.

In internal combustion engines, a portion of liquid fuel spray may directly land on the liner and mix with oil (lubricant), forming a fuel-oil film (~10μm) that is much thicker than the original oil film (~0.1μm). When the piston retracts in the compression stroke, the fuel-oil mixture may have not been fully vaporized and can be scraped by the top ring into the 1st land crevice and eventually enter the combustion chamber in the format of droplets. Studies have shown that this mechanism is possibly a leading cause for low-speed pre-ignition (LSPI) as the droplets contain oil that has a much lower self-ignition temperature than pure fuel. In this interest, this work aims to study the oil-fuel interactions on the liner during an engine cycle, addressing molecular diffusion (in the liquid film) and vaporization (at the liquid-gas interface) to quantify the amount of fuel and oil that are subject to scraping by the top ring, thereby exploring their implications on LSPI and friction.

This paper provides an overview of spark-ignition engine unburned hydrocarbon emissions mechanisms, and then uses this framework to relate measured engine-out hydrocarbon emission levels to the processes within the engine from which they result. Typically, spark-ignition engine-out HC levels are 1.5 to 2 percent of the gasoline fuel flow into the engine; about half this amount is unburned fuel and half is partially reacted fuel components. The different mechanisms by which hydrocarbons in the gasoline escape burning during the normal engine combustion process are described and approximately quantified. The in-cylinder oxidation of these HC during the expansion and exhaust processes, the fraction which exit the cylinder, and the fraction oxidized in the exhaust port and manifold are also estimated.

The combustion behavior in a modem 4-valve engine using a broad range of methanol/gasoline fuel mixtures was studied. The initial flame development was examined by using a spark plug fiber optics probe. Approximately, the kernel expansion speed, Sg, is relatively unchanged from M0 to M40; jumps by ∼30% from M40 to M60; and then remains roughly constant from M60 to M100. Statistics of the IMEP indicate that at a lean idle condition the combustion rate and robustness correlate with Sg: a higher value of Sg gives better combustion. Thus M60-M100 fuels give better idle combustion behavior than the M0-M40 fuels.

A method for determining the flame contour based on the flame arrival time at the fiber optic (FO) spark plug and at the head gasket ionization probe (IP) locations has been developed. The experimental data were generated in a single-cylinder Ricardo Hydra spark-ignition engine. The head gasket IP, constructed from a double-sided copper-clad circuit board, detects the flame arrival time at eight equally spaced locations at the top of the cylinder liner. Three other IP's were also installed in the cylinder head to provide additional intermediate data on flame location and arrival time. The FO spark plug consists of a standard spark plug with eight symmetrically spaced optical fibers located in the ground casing of the plug. The cylinder pressure was recorded simultaneously with the eleven IP signals and the eight optical signals using a high-speed PC-based data acquisition system.

This paper uses a model which calculates the flame kernel formation and its early development in spark ignition engines to examine the causes of cycle-to-cycle combustion variations. The model takes into account the primary physical factors influencing flame development. The spark-generated flame kernel size and temperature required to initialize the computation are completely determined by the breakdown energy and the heat conduction from burned region to unburned region. In order to verify the model, the computation results are compared with high-speed Schlieren photography flame development data from an operating spark-ignition engine; they match remarkably well with each other at all test conditions. For the application of this model to the study of cycle-to-cycle variation of the early stage of combustion, additional input is required.

A model has been developed which predicts engine-out hydrocarbon (HC) emissions for spark-ignition engines. The model consists of a set of scaling laws that describe the individual processes that contribute to HC emissions. The model inputs are the critical engine design and operating variables. This set of individual process scaling relations was then calibrated using production spark-ignition engine data at a fixed light-load operating point. The data base consisted of engine-out HC emissions from two-valve and four-valve engine designs with variations in spark timing, valve timing, coolant temperature, crevice volume, and EGR, for five different engines. The model was calibrated separately for the three different engines to accommodate differences in engine design details and to determine the relative magnitudes of each of the major sources. A good fit to this database was obtained.

The fuel injection process in the port of a firing 4-valve SI engine at part load and 25°C head temperature was observed by a high speed video camera. Fuel was injected when the valve was closed. The reverse blow-down flow when the intake valve opens has been identified as an important factor in the mixture preparation process because it not only alters the thermal environment of the intake port, but also strip-atomizes the liquid film at the vicinity of the intake valve and carries the droplets away from the engine. In a series of “fuel-on” experiments, the fuel injected in the current cycle was observed to influence the fuel delivery to the engine in the subsequent cycles.

A Lagrangian random vortex-boundary element method has been developed for the simulation of unsteady incompressible flow inside three-dimensional domains with time-dependent boundaries, similar to IC engines. The solution method is entirely grid-free in the fluid domain and eliminates the difficult task of volumetric meshing of the complex engine geometry. Furthermore, due to the Lagrangian evaluation of the convective processes, numerical viscosity is virtually removed; thus permitting the direct simulation of flow at high Reynolds numbers. In this paper, a brief description of the numerical methodology is given, followed by an example of induction flow in an off-centered port-cylinder assembly with a harmonically driven piston and a valve seat situated directly below the port. The predicted flow is shown to resemble the flow visualization results of a laboratory experiment, despite the crude approximation used to represent the geometry.

This paper describes the use of a modified quasi-dimensional spark-ignition engine simulation code to predict the extent of cycle-to-cycle variations in combustion. The modifications primarily relate to the combustion model and include the following: 1. A flame kernel model was developed and implemented to avoid choosing the initial flame size and temperature arbitrarily. 2. Instead of the usual assumption of the flame being spherical, ellipsoidal flame shapes are permitted in the model when the gas velocity in the vicinity of the spark plug during kernel development is high. Changes in flame shape influence the flame front area and the interaction of the enflamed volume with the combustion chamber walls. 3. The flame center shifts due to convection by the gas flow in the cylinder. This influences the flame front area through the interaction between the enflamed volume and the combustion chamber walls. 4. Turbulence intensity is not uniform in cylinder, and varies cycle-to-cycle.

The occurrence of liquid fuel in the cylinder of automotive internal combustion engines is believed to be an important source of exhaust hydrocarbon (HC) emissions, especially during the warm-up process following an engine start up. In this study a Phase Doppler Particle Analyzer (PDPA) has been used in a transparent flow visualization combustion engine in order to investigate the phenomena which govern the transport of liquid fuel into the cylinder during a simulated engine start up process. Using indolene fuel, the engine was started up from room temperature and run for 90 sec on each start up simulation. The size and velocity of the liquid fuel droplets entering the cylinder were measured as a function of time and crank angle position during these start up processes. The square-piston transparent engine used gave full optical access to the cylinder head region, so that these droplet characteristics could be measured in the immediate vicinity of the intake valve.

Smoke is emitted in diesel engines because fuel injected into the combustion chamber burns with insufficient oxygen. The emission smoke from diesel engines is a very important air pollution problem. Smoke emission, which is believed to be largely related to the diffusion combustion in diesel engines, results from pyrolysis of fuel not mixed with air. Therefore, the smoke emission is dependent on diffusion combustion phenomena, which are controlled by engine parameters. This paper presents an analysis of combustion by relating the smoke emission with heat release in diesel engines. An analysis is made of the diffusion combustion quantity, the smoke emission, and the fraction of diffusion combustion as related to the engine parameters which are air-fuel ratio, injection timing, and engine speed.

Cranking and startup fuel control has become increasingly important due to ever tightening emission requirements. Additionally, engine-off strategies during idle will require substantially more engine startup events with the associated need for very clean starts. Thus, knowledge of an engine's Air-Fuel Ratio (AFR) during its early cycles is necessary in order to optimize cranking and startup fueling. This paper examines and compares two methods of measuring an engine's AFR during engine startup (approximately the first second of operation); an in-cylinder technique using a Fast Flame Ionization Detector (FFID) and the conventional exhaust based Universal Exhaust Gas Oxygen (UEGO) sensor method. Engine starts using a Ford Zetec engine were performed at three different temperatures (0, 20 and 90 C) as well as different initial engine starting positions.