Optical Access Engine Research Articles

Advanced Flame front Detection in Combustion Processes Using Autoencoder Approach

This research explores the detection of flame front evolution in spark-ignition engines using an innovative neural network, the autoencoder. High-speed camera images from an optical access engine were analyzed under different air excess coefficient λ conditions to evaluate the autoencoder’s performance. This study compared this new approach (AE) with an established method used by the same research group (BR) across multiple combustion cycles. Results revealed that the AE method outperformed the BR in accurately identifying flame pixels and significantly reducing overestimations outside the flame boundary. AE exhibited higher sensitivity levels, indicating its superior ability to identify pixels and minimize errors compared to the BR method. Additionally, AE’s accuracy in representing combustion evolution was notably improved, offering a more detailed depiction of the process. AE’s strength lies in its independence from specific threshold searches, a requirement in the BR method. By relying on learned representations within its latent space, AE eliminates laborious threshold exploration, ensuring reliability and reducing workload pressures. Comparative analyses consistently confirmed AE’s superior performance in accurately reproducing and delineating combustion evolution compared to BR. This study highlights AE’s potential as a promising technique for precise flame front detection in combustion processes. Its ability to autonomously extract features, minimize errors, and enhance overall accuracy signifies a significant step forward in analyzing flame fronts. AE’s reliability, reduced need for manual intervention, and adaptability across various conditions suggest a promising future for improving combustion analysis techniques in spark-ignition engines with optical access.

Comparative analysis of thermal and non-thermal discharge modes on ultra-lean mixtures in an optically accessible engine equipped with a corona ignition system

Radio-Frequency corona-based ignition systems are currently being investigated by the engine research community for their capability to ensure robust combustion at challenging operating conditions such as lean mixture and/or high exhaust gas recirculation (EGR) dilution. The low-temperature plasma (LTP) produced by the corona discharge accelerates the early flame development thanks to the production of combustion precursors like radical and excited species. Plasma benefits can be maintained with extensive management of the control parameters, preventing the transition from the non-thermal discharge mode to a thermal one.However, under ultra-lean conditions, where the ignition probability significantly decreases, the LTP discharge may not be suitable for promoting the combustion process because of insufficient thermal energy transfer and scarcity of available fuel, which is essential for radical production. Therefore, it may be necessary to increase the volume in which the plasma interacts to ensure robust ignitions and stable flame propagations. For this reason, the present work aims to assess and quantify the differences, in terms of combustion performance, between a corona igniter operating at LTP mode and with electric arcs occurring, at ultra-lean operating conditions. In arc-mode, the igniter produces volumetric discharges, facilitating extensive mixing of the mixture and resulting in advantageous effects on kernel formation and propagation. Tests were carried out in a single-cylinder optical access engine under low load and low speed, using gasoline as fuel. Close to the leanest stable limit found, the LTP mode shows a lower cycle-to-cycle variability. Above such a threshold, the thermal discharge can guarantee stable combustion events while the LTP discharge leads to unstable working conditions. Indicating, imaging, and pollutant emission analysis are carried out to evaluate the effects of thermal and non-thermal plasma modes on combustion characteristics and their environmental impact.

Using a Machine Learning Approach to Evaluate the NOx Emissions in a Spark-Ignition Optical Engine

Currently, machine learning (ML) technologies are widely employed in the automotive field for determining physical quantities thanks to their ability to ensure lower computational costs and faster operations than traditional methods. Within this context, the present work shows the outcomes of forecasting activities on the prediction of pollutant emissions from engines using an artificial neural network technique. Tests on an optical access engine were conducted under lean mixture conditions, which is the direction in which automotive research is developing to meet the ever-stricter regulations on pollutant emissions. A NARX architecture was utilized to estimate the engine’s nitrogen oxide emissions starting from in-cylinder pressure data and images of the flame front evolution recorded by a high-speed camera and elaborated through a Mask R-CNN technique. Based on the obtained results, the methodology’s applicability to real situations, such as metal engines, was assessed using a sensitivity analysis presented in the second part of the work, which helped identify and quantify the most important input parameters for the nitrogen oxide forecast.

An Investigation of the Impact of Combustion Chamber Geometry on Turbulent Burning Speeds in a Thermodynamic Model

Abstract Combustion is the main parameter that affects efficiency and exhaust gas emissions. Recently, different studies have been carried out to increase the combustion rates due to the increasing use of the alternative fuels and lean mixtures in spark ignition engines. In general, in the absence of systems such as an optical access engine or ionization probes, combustion process evaluation is done based on cylinder pressure. In this study, the effect of different combustion chamber geometries on the turbulent burning speeds was investigated experimentally and theoretically. A three-zone, quasi-dimensional thermodynamic model for a spark ignition engine was constructed. Measured cylinder pressure data were used to establish the transition between the zones of the model and determine turbulent burning speeds. Two different turbulent speeds were calculated in the model, these are flame propagation and consumption speeds. It was seen that combustion chamber geometry significantly affected turbulent burning speeds. In MAN-Ricardo shapes and cylindrical shapes, the turbulent burning speeds decreased after it reached a maximum value in the combustion period. In flat geometry, without any bowl, speed continuously decreased different from other two designs. By means of a quasi-dimensional thermodynamic model, mean values of the turbulent burning and the flame propagation speeds can be calculated without having any optical observation.

Modeling of thermal and kinetic processes in non-equilibrium plasma ignition applied to a lean combustion engine

In recent years novel ignition systems have been developed to enable stable and efficient engine operations with lean mixtures. Among them, radio-frequency corona ignition systems create discharges that involve a much wider region compared to traditional spark, and produce non equilibrium plasma with high levels of active radicals and excited species. These devices considerably increase the early flame growth speed and extend stable operating limits. With the aim of expanding the knowledge on high efficiency lean-burn SI engines, this paper investigates and compares the combustion development generated by spark and corona ignitions through computational fluid dynamics, within the Reynolds–Averaged Navier–Stokes framework for turbulence modeling. In order to simultaneously take thermal and chemical effects into account, the Perfectly Stirred Reactor combustion model is used. Experimental data are also collected for validation in an optical access engine, for different mixture levels, from stoichiometric to very lean. The faster burn rate generated by the corona system in the initial stage of the combustion is well predicted by the simulations, in all the relative air–fuel ratio conditions. Remarkably, as the mixture becomes lean, simulations are able to capture the non-linear transition from fast to slow kernel growth, before a self-sustainable flame propagation is established. This correlates very well with the measured engine cyclic variability and the corresponding steep change in the duration of the flame kernel formation. Ultimately, this study highlights the important role of the atomic oxygen, as active radical, in promoting and enhancing the combustion initiated by a corona discharge, in addition to the volumetric ignition effect. By contrast, the validated simulations allow to explain that the high-temperature thermal plasma generated in a traditional spark discharge is insensitive to kinetic aspects.

LES investigation of cycle-to-cycle variation in a SI optical access engine using TFM-AMR combustion model

Multi-cycle large-eddy simulations (LES) are performed to investigate combustion cycle-to-cycle variability (CCV) in a gasoline spark ignited optical access engine operating under homogeneous stoichiometric conditions. Combustion is addressed with the Thickened Flame Model (TFM) and finite rate chemistry is accounted for through a reduced oxidation reaction mechanism. In view of the fact that computational costs of LES engine simulations are still very high today, this work investigates the use of adaptive mesh refinement (AMR) in the flame zone in conjunction with the artificial flame thickening applied by the TFM model. The paper discusses how the resulting coupled TFM-AMR combustion model allows good resolution of the flame, maintaining accuracy at acceptable costs. First, the details of the coupled model are presented and the effects of the parameters are explored, highlighting their impact on the combustion prediction. Then, computational fluid dynamics (CFD) simulation results are validated against experimental data collected in a low-speed low-load engine point, by comparing 20 LES cycles and 100 measured cycles, for mass fraction burned, combustion phasing, flame images and CCV indices. Lastly, a detailed investigation on the fastest and slowest numerical cycles is presented, analyzing instantaneous flame structures, ignition behaviors, propagation speeds, and probability density function (PDF) of the instantaneous velocity fluctuation around the spark region. The results show that combustion variability is highly correlated to the resolved velocity field and the resolved turbulence intensity, which is found to be the main cause of CCV and affects the early flame kernel growth. This work is an early attempt to use TFM-AMR combustion model for LES simulations of internal combustion engines.

Streamers Variability Investigation of a Radio-Frequency Corona Discharge in an Optical Access Engine at Different Speeds and Loads

Engine research community interest in the Radio-Frequency corona-based ignition systems is currently gaining in importance mainly due to their capability to ensure robust combustion at challenging operating conditions such as very lean mixture and/or high EGR dilution. The benefits of Corona low-temperature plasma foster the early flame development thanks to combustion precursors production and to a more energetic and volumetric discharge, resulting in a larger amount of involved mixture. The corona discharge generates ionizing waves, named streamers, whose temporal and spatial variability in orientation, length and branching can affect the combustion onset and, therefore, the engine cycle-to-cycle variability. In this work, the discharge natural luminosity of a RF corona igniter, characterized by four tips electrodes, was recorded in an optically accessible engine via high-speed camera detection. A preliminary statistical analysis of the spatial and temporal streamer variability was performed by operating in motored conditions. Four different engine speeds and two different loads were explored in order to deeply investigate the streamer behaviour at diverse engine operating conditions. A comparison between a motored and a lean operating condition is also proposed to analyse, at a specific engine speed, the mixture influence on the streamers propagation before the start of the combustion.

The Role of Multiple Injections on Combustion in a Light-Duty PPC Engine

This paper presents a numerical investigation of the ignition and combustion process of a primary reference fuel in a partially premixed light-duty internal combustion (PPC) engine. Partially pre-mixed combustion is achieved by employing a multiple injection strategy with three short injection events of fuel pulses. The timing of the first two fuel pulses, 48 and 22 crank angle degrees before top dead center, are chosen with the purpose to stratify the fuel and air charge, whereas the third injection, at five crank angle degrees before top dead center, serves as an actuator of the main heat release. In addition to this baseline injection, three alternative injection strategies are studied, including a split-fuel two-injection strategy and modified triple-injection strategies. Large eddy simulations are employed utilizing a skeletal chemical kinetic mechanism for primary reference fuel capable of capturing the low-temperature ignition and the high temperature combustion. The large eddy simulation (LES) results are compared with experiments in an optical accessible engine. The results indicate that the first ignition sites are in the bowl region where the temperature is relatively higher, and the reaction fronts thereafter propagate in the swirl direction and towards the centerline of the cylinder. The charge from the first two injections initially undergoes low-temperature reactions and thereafter high-temperature reservoirs are formed in the bowl region. The main heat-release is initiated in the engine when the fuel from the third injection reaches the high-temperature reservoirs. Finally, the remaining fuel in the lean mixtures from the first two injections is oxidized. By variation of the injection strategy, two trends are identified: (1) by removing the second injection a higher intake temperature is required to enable the ignition of the charge, and (2) by retarding second injection, a longer ignition delay is identified. Both can be explained by the stratification of fuel and air mixture, and the resulting reactivity in various equivalence ratio and temperature ranges. The LES results reveal the details of the charge stratification and the subsequent heat release process. The present results indicate a rather high sensitivity of partially premixed combustion process to the injection strategies.

Multi-Plane PIV Measurements in a Gasoline Direct Injection Engine

<div class="section abstract"><div class="htmlview paragraph">The flows in-cylinder have a profound effect on the mixture preparation and subsequent combustion in all engines. These flows are highly three-dimensional in nature and information from multiple planes is required to characterise the flow dynamics. The flow measurements reported here are from three orthogonal planes in an optical access engine that is based on the Jaguar Land Rover AJ200 Gasoline Direct Injection (GDI) engine. Particle Image Velocimetry (PIV) measurements have been taken every 5°CA from the start of induction to the end of compression. Data have been obtained from 300 cycles for separate experiments measuring flows in the tumble plane, the swirl plane and the cross-tumble plane. Vector comparison metrics are used to quantitatively compare ensemble averaged PIV flow fields to Computational Fluid Dynamics (CFD) simulations across each plane in terms of both the velocity magnitude and direction. These quantitative metrics enable identification of regions of interest and differences in flow characteristics across a range of engine operating conditions. Combination of the PIV vector fields from all three planes into composite flow fields enables the interaction between the flows in the three planes to be visualised. During induction the intake jet is visible in all three planes, while the swirl plane measurements show a pair of counter-rotating vortices either side of the intake jet and the tumble plane shows the development of the tumble flow. During compression the center of the tumble vortex traverses the cylinder leading to flow reversal in the swirl and cross tumble measurement planes.</div></div>

Combustion CFD Modeling of a Spark Ignited Optical Access Engine Fueled with Gasoline and Ethanol

In this study we present the Computational Fluid Dynamics (CFD) modeling of the combustion process using detailed chemistry in a spark-ignited (SI) optical access engine operated at part load using gasoline and ethanol as fuels. Simulation results are compared against experimental optical and indicating data. The engine is installed at the Department of Engineering of the University of Perugia, and it features a four-valve head, a transparent flat piston and a port-fuel-injection (PFI) system. Full open cycle simulations have been performed using the commercial code CONVERGE. The combustion process has been simulated using detailed chemistry and adaptive mesh refinement (AMR) to resolve in detail and track the reaction zone, in a Reynolds Averaged Navier-Stokes (RANS) modeling framework. In-cylinder pressure, heat release, and flame morphology have been compared with experimental indicating and imaging data. Tests and simulations span different air-fuel ratios in lean and rich conditions (relative air-fuel ratio λranges from 0.9 to 1.1). Results indicate that simulations are able to predict experimental data with high accuracy. Variations due to changing fuel type and air-fuel ratio are well captured. The computational cost to achieve grid-independent results has been evaluated and it is also not excessively high. Taking into account that the engine speed was quite low, i.e., 900rpm, we conclude that, in this condition, detailed chemistry coupled with RANS works satisfactorily without turbulence chemistry interaction sub-models, and therefore without any tunings.

Investigation on the flame characteristics in a PCCI engine using endoscope system with micro-cassegrain

Diesel engines are most frequently used in vehicles because of their high efficiency and durability. However, nitrogen oxide (NOx) and particulate matter (PM) from diesel engines are became problem. Recently, the premixed charged compression ignition (PCCI) technique has been studied specifically to reduce NOx and PM emissions simultaneously. While the PCCI engine has the advantage of reducing NOx and PM emissions, control of the combustion phase is difficult. Therefore, it is very important to understand the flame developing process and heat release pattern in a real PCCI engine. However, most research into the characteristics of PCCI combustion has used an optical access engine. This paper describes the combustion characteristics obtained using a micro-Cassegrain sensor and discusses how the sensor can be used to directly detect the flame characteristics for future control of PCCI combustion. From this research, the differences of flame characteristics between PCCI engine and conventional diesel engine were clarified.

Fluid-dynamic and numerical aspects in the simulation of direct CNG injection in spark-ignition engines

This paper presents a detailed discussion on the numerical simulation of the underexpanded gas efflux from an outward-opening poppet-valve injector into an engine combustion chamber. The aim of the paper is to optimize the numerical simulation strategy for direct gas injection, in view of its application to internal combustion (IC) engines. In the first part of the paper, the widely studied case of a two-dimensional compressible flow is examined, and the main guidelines for the development of an effective numerical model for compressed natural gas (CNG) direct injection simulation are given, with specific reference to IC engines. The second part of the paper is devoted to the description of the numerical model developed and validated by the authors within the Star-CD environment, which is characterized by the presence of two distinct meshes. The first is built manually and covers the region surrounding the injector exit, whereas the second one covers most of the engine chamber and is built using the Es-ICE tool. A careful grid-independence study has been carried out in both the first and second part of the paper, and the influence of the spatial discretization of the convective fluxes has been discussed as well.The analyses have shown that a resolution of 40 cells in the nozzle height should be adopted to describe the typical phenomena that characterize an underexpanded free jet, unless a second order scheme can be implemented. However, as far as the simulation of the jet penetration time-history and its mixing with the surrounding air is concerned, sufficiently accurate results can also be obtained by using 20 cells per nozzle diameter and the first-order upwind scheme. As for the direct injection engine model, 16 cells across the nozzle lift represent a good compromise between accuracy and reliability of the results and the required computational time. The model has been validated with the support of experimental PLIF images in an optical-access engine, and has shown overall good accuracy and reliability, thus suggesting it is suitable for mixture formation analysis.

Combustion Analysis in an Optical Access Engine

The ever more strict pollutant emissions regulations urge to make further steps in the evolution of the internal combustion engines, in particular as far as the emissions generation. A direct observation of the phaenomena taking place inside the cylinder can hence help in better understanding the effects obtainable by using different solutions (injection systems, ignition systems,…).To this end, the use of an optical access engine can be very helpful. This paper presents the first step made at the Department of Industrial Engineering of the University of Perugia for the set-up of an experimental line to be used for the analysis of combustion events observed in a single cylinder optical access engine.The combustion was characterized by using different techniques: thermodynamic (heat release) based on in-cylinder pressure measurements, and non intrusive optical observations by means of images acquisition. The flame images collected by a synchronized CCD camera were post-processed in order to evaluate the rates of the in-cylinder flame front evolution, observing the combustion process from the flame kernel formation on.Different fuels were used in the spark ignition, four valves, optical access engine: pure gasoline, pure ethanol, a 50% gasoline/ethanol blend. The experimental tests were carried out at 900rpm engine speed, in rich and lean conditions, the latter being of interest for the achievement of significant CO2 reduction.

Lift-off and stabilization of n-heptane combustion in a diesel engine with a multiple-nozzle injection

This paper presents a joint numerical and experimental investigation of flame lift-off and stabilization mechanisms in heavy-duty diesel engines. The injection strategy, employing different nozzle configurations, allows for quantification of the impact of varying inter-jet angle spacing in the presence of swirl. For this purpose, three different inter-jet angles are chosen in this study; 45°, 90° and 135°. Large-eddy simulations are performed utilizing a detailed chemical kinetic mechanism for n-heptane to resolve the turbulent fuel and air mixing and to capture the important species surrounding the ignition and flame-fronts to describe the flame stabilization process. Measurements are carried out for OH chemiluminescence to identify the flame lift-off position in an optical accessible engine. In general, the swirl flow in the ambient air shows a great impact on the lift-off, with a 15% difference in the lift-off lengths on the upwind and downwind side of the jet. The LES results show that important ignition reactions undergo in a broad region in front of the lift-off position. With decreasing inter-jet angle, it is shown that the impact of transportation of hot products from adjacent jets becomes more prominent. Hot reservoirs surrounding the lift-off length increase the local ambient temperature and augment the auto-ignition process by mixing of the cold injected fuel and hot air.

A study of mixture preparation and PM emissions using a direct injection engine fuelled with stoichiometric gasoline/ethanol blends

The effect of gasoline/ethanol blends in different blending proportions (E0, E10, E20, E50, E70, E85 as #% by volume) on the characteristics of size-resolved particulate number and mass concentrations was investigated in a single-cylinder optical access engine using a differential mobility spectrometer (DMS500) under cold and warm conditions (20°C, 80°C) at a stoichiometric condition (1500rpm, 0.5bar manifold absolute pressure). The effects on the fuel spray characteristics were also investigated by capturing the injection images and by post-processing to compare the temporal development of the spray with the different fuels. In-cylinder fFID (fast Flame Ionisation Detector) measurements were undertaken to assess the mixture in-homogeneity in the region of the spark plug. As the ethanol volumetric percentage increases, both the total Pn (Particulate number) and Pm (Particulate mass) increase by a maximum of 16 and 11 for cold conditions and 7 and 8 for warm conditions. This is in agreement with the natural flame chemiluminescence images which showed that fuels with high ethanol proportions generated more soot than fuels with small blending percentages of ethanol. Spray images and combustion analysis showed that ethanol addition results in protracted injection and combustion periods. The effect on spray characteristics is more profound under cold conditions than under warm conditions. The mixture in-homogeneity increases as the ethanol content increases for E50, E70, and E85 and this would explain the increased level of particulate emissions, although the trend is less clear for the lower ethanol content fuels.

MD2-5 An Investigation of DME HCCI Combustion Using Spectroscopic Analysis(MD: Measurement and Diagnostics,General Session Papers)

In order to control the ignition timing and overall combustion rate of HCCI, it is important to have a firm understanding of the chemical reactions of fuel because HCCI can be characterized as controlled chemical auto-ignition process. One technique that can be employed to gain a better understanding of the HCCI combustion process is chemiluminescence, which is the light emission from the excited state of intermediates or radicals. The objective of this study is to experimentally investigate the correlation between chemiluminescence of the active species, which play important role in the ignition, and HCCI combustion characteristics for providing the way to control auto-ignition process. In addition, a combination of experimental and computational results is employed. The experiments were conducted using optical accessible engine, photo-multiplier and band-pass filters. Senkin application of the CHEMKIN-II kinetics rate code was used for computation. As the result, there was a precise correlation between the rate of heat release and chemiluminescence intensity of OH (307.4 nm) and all band (350-700 nm) by tracing the in-cylinder mass-averaged. The start timing of emitting luminescence in OH coincided with H_2O_2 dissociation, and the peak timing of emitting luminescence in all band corresponded to the CO-O recombination spectrum.

G070044 PIV法によるブローダウン過給ガソリンHCCI機関の筒内流動の可視化([G07004]エンジンシステム部門一般セッション(4):エンジンシステム)

The objective of this research is to investigate the effect of the EGR guide employed to generate a strong in-cylinder thermal stratification on the recharged EGR gas flow into the cylinder in the gasoline HCCI engine equipped with the blowdown supercharging (BDSC) system. The recharged EGR gas flow was visualized using an optical access engine. A steady gas flow from the exhaust port into the cylinder was visualized via PIV method. The effect of the pressure difference between the exhaust port and the cylinder and the geometry of the EGR guide on the recharged gas flow were investigated. The results indicate that the flow field pattern is largely affected by the EGR guide height, and that an increase in the EGR guide height is effective to generate a strong in-cylinder thermal stratification.

An Investigation of Unburned Hydrocarbon Emissions in Wall Guided, Low Temperature Diesel Combustion

The formation mechanisms of unburned hydrocarbons (HC) in low NOx, homogeneous type Diesel combustion have been investigated in both standard and optical access single cylinder engines operating under low load (2 and 4 bar IMEP) conditions. In the standard (i.e. nonoptical) engine, parameters such as injection timing, intake temperature and global equivalence ratio were varied in order to analyse the role of bulk quenching on HC emissions formation. Laser-Induced Fluorescence (LIF) imaging of in-cylinder unburned HC within the bulk gases was performed on the optical-access engine. Furthermore, studies were performed in order to ascertain whether the piston topland crevice volume contributes significantly to engine-out HC emissions. Finally, the role of piston-top fuel films and their impact on HC emissions was studied. This was investigated on the all-metal engine using two fuels of different volatilities. Parallel studies were also performed on the optical-access engine via in-cylinder tracer Laser-Induced Fluorescence (LIF) imaging. Results obtained in the standard and optical access engines revealed that bulk quenching represents one of the most significant sources of unburned HC for the wall guided combustion chamber geometry. Bulk quenching occurs as a result of incomplete fuel oxidation reactions in regions where the local equivalence ratio is either too fuel-lean or too fuel-rich or alternatively in excessively low temperature zones within the combustion chamber. Experimental data obtained from both the standard and optical access engines also revealed that liquid film formation occurs, and is particularly prevalent for early Start Of Injection (SOI) strategies. Furthermore, liquid films remain present at the end of combustion and are believed to represent a significant source of engine-out HC emissions. In-cylinder imaging of liquid films suggest that the film eventually detaches from the piston surface later during the expansion stroke, resembling a flash boiling phenomenon. The results appear to confirm that unburned fuel arising from piston-top fuel films contribute directly to the engine-out HC emissions.

Full-bore crank-angle resolved imaging of combustion in a four-stroke gasoline direct injection engine

The conventional Bowditch piston arrangement in an optical access engine limits the field of view to about half the bore area. By using a transparent crown (with a flat top) and incorporating a concave surface into the underside of the piston, then the piston crown acts as a diverging lens. Appropriate choice of the radius of curvature and the piston crown thickness provides an image of the full bore. There is of course optical distortion in the image but, since the piston position is known for each image, then it is comparatively simple to reconstruct undistorted images and videos. As an illustration of the technique, results are presented here for flame front tracking, and estimates of the combustion temperature and soot loading by a colour ratio pyrometry technique. This full-bore imaging technique is also applicable to many other optical diagnostic techniques. For a motored engine with particle imaging velocimetry and an optical cylinder liner, then the light sheet could enter either through the piston to illuminate a vertical plane or through the cylinder liner and be imaged through the piston. The choice of laser-based combustion diagnostic techniques would be limited by the transmission range of the piston window.

Diesel nozzle geometry influence on spray liquid-phase fuel penetration in evaporative conditions

An experimental study of real multi-hole Diesel nozzles is performed under current DI Diesel engines operating conditions. The aim of the investigation is to study the influence of orifice geometry on the flow at the nozzle exit and to analyse its effect on the spray in evaporative conditions. Special attention is taken in the study of the influence of cavitation on the orifice internal flow and spray development. The spray liquid-phase fuel penetration has been characterized. The visualization was made in a wide optical access engine for different operating conditions. From the measurements, the dependencies of nozzle geometry, injection conditions and ambient conditions on liquid-phase length were studied and analyzed. A model for liquid-phase fuel penetration in diesel sprays based on nozzle flow parameters has been proposed and validated.