Experimental and numerical investigations on the mechanisms leading to the accumulation of particulate matter in lubricant oil.
The accumulation of particulate matter in lubricant oil can become an important issue in Diesel engines where large amounts of Exhaust Gas Recirculation (EGR) are used at medium to high load operating conditions. Indeed, the transport and subsequent accumulation of particulate matter in the engine oil can negatively impact the oil lubricant properties which is critical to ensure mechanical durability and limit the vehicle Total Cost of Ownership (TCO) by reducing the servicing intervals. The objective of this investigation was to gain an improved understanding of the underlying mechanisms that are responsible for the accumulation of particulate matter in the lubricating oil, and ultimately provide design guidelines to help limit this phenomenon.
The present study presents the development and validation of experimental and numerical tools used to investigate this phenomenon. Several advanced diagnostic techniques were developed and applied on an optically-accessible single cylinder Diesel engine to detect the presence of particulates and quantify their concentration in two particular zones: (1) in the upper part of the cylinder liner where particulate matter is believed to be absorbed into the oil film and (2) in the engine blow-by gases where particulates can be transported via the piston ring-pack to the oil sump reservoir. The accumulation of soot particulates on the surface of the upper part of the cylinder liner was characterized using a modified Laser Extinction Method (LEM), while the concentration of particulates in the engine blow-by gases was measured using a DMS500 soot sensor. In parallel to the experimental study, 3D numerical computations were performed and provided additional information to the experimental results. In the present paper, the various experimental and numerical methods are presented and first validation results are discussed.
CITATION: Laget, O., Malbec, L., Kashdan, J., Dronniou, N. et al., "Experimental and Numerical Investigations on the Mechanisms Leading to the Accumulation of Particulate Matter in Lubricant Oil," SAE Int. J. Engines 9(4):2016, doi:10.4271/2016-01-2182.
Prior investigations performed on various heavy duty Diesel engines show that combustion strategies where retarded injection timing and high EGR ratios are employed with the aim of achieving vehicle exhaust emissions compliance can result in contamination of the lubricant oil by soot particles, up to levels of about 10% in weight . This accumulation is particularly critical on engine operating points where high smoke levels are observed such as at mid to high loads (IMEP>12bar), where high EGR rates are employed or during regeneration strategies for exhaust gas catalysts.
The increased presence of soot in the crankcase oil is believed to aggravate engine wear . This affects mainly the engine valve train, piston rings and cylinder liner wear  and imposes more frequent servicing intervals in order to prevent a reduction in engine performance or, in extreme cases, engine failure. Accumulation of particulate matter in the lubricant oil therefore represents an interest for optimizing both engine reliability and Total Cost of Ownership (TCO). Even if an increase in engine wear which according to  is dependent on oil soot content and soot particle size, can potentially be limited by using specific oil formulations (base oil and additives) [2, 3], there is still a clear need to better understand the physical mechanisms explaining the accumulation of particulate matter in lubricant oil.
Different pathways can explain the accumulation of soot in oil. They can either be transported from the combustion chamber to the crankcase within the blow-by gases, or they can be deposited directly on the lubricant oil film on the liner and then subsequently scraped by the piston rings.
Soot deposition on the liner can be explained by several mechanisms, described in : thermophoresis, brownian diffusion, turbulent diffusion, inertial impingement, electrophoresis or gravitational sedimentation, but the principal mechanism seems to be thermophoresis . One previous study claims that the deposit mechanism accounts for more than 97% of the soot present in engine oil , but the relative weight of each of these mechanisms is still not well known.
Regarding potential solutions to this issue, a recent study shows a combustion chamber shape evolution aimed at improving combustion efficiency while reducing levels of in-cylinder soot and in particular the soot-liner interaction .
Based on the literature review, the two main mechanisms explaining the presence of soot in oil are the following:
* Soot brought into contact with the cylinder liner is absorbed by the oil film and subsequently scraped by the piston rings.
* The soot located in the vicinity of the piston top land zone is entrained directly by the blow-by gases into the engine crank case and subsequently absorbed in the oil sump.
The aim of this study was to gain a better understanding of the processes controlling soot accumulation in the lubricant oil, in order to subsequently identify new combustion chamber designs and/or combustion strategies which could help limit this phenomenon. This study focused first and foremost on establishing whether the two mechanisms were indeed responsible for the accumulation of particulate matter in the lubricant oil. Secondly, the influence of various engine parameters was also investigated such as Start of Injection (SOI) timing, EGR rate and engine load (IMEP). This requires the ability to measure or estimate:
* The quantity of soot present in the combustion chamber;
* The quantity of soot present in the blow-by gases;
* The quantity of soot deposited on the cylinder liner.
This paper presents experimental and numerical methodologies that were developed specifically for this study and preliminary results. A particular focus is given to the validation of the developed methodologies which enabled measurements or numerical predictions of these 3 quantities, together with their temporal evolution either during one engine cycle and/or during several consecutive engine cycles. The retained approach consisted in combining experimental (optical engine) and numerical (3D CFD) tools. Experimental results were used both for an in-depth analysis of the different physical phenomena and for the construction of a database which served to validate the 3D CFD simulations. Once validated, numerical simulations were performed and provided access to parameters that were not accessible by experimental measurements. Finally 3D CFD simulations were also used to perform variations and optimization of combustion chamber design
The advanced diagnostics are detailed in the following experimental setup section. The numerical tool is based on 3D CFD RANS computations. During this study specific data post-processings were developed, they are exposed in the "Numerical setup" section along with the numerical models and the main assumptions used. Finally, a first set of results is presented, with a first attempt to propose a mechanism explaining the accumulation of particulate matter in the lubricant oil.
This study was performed within the framework of the Groupement Scientifique Moteur (GSM), a collaborative R&D program between IFPEN, and the French car manufacturers Peugeot-Citroen (PSA group) and Renault (RSA).
An optical-access single cylinder engine was used in this study with a 0,5 liter displacement and is representative of a light duty direct-injection (DI) 4-stroke Diesel engine. A schematic view of this engine is presented in Figure 2, and the general engine characteristics are listed in Table 1.
This engine can be equipped in several optical configurations to enable the application of laser based diagnostics for in-situ measurements:
* Fully/partially optical crown: this corresponds to the upper part of the liner, in which transparent windows can be installed. This optical crown can also be fully optical made of sapphire in order to have a full optical access to the upper part of the combustion chamber.
* Replacement of one of the two exhaust valves by a window providing a view from the cylinder head down into the combustion chamber;
* Optical piston: an optical bowl window is used;
* Optical liner: the entire liner is transparent.
There are several differences between optical and all-metal engines that have been studied in . Given the context of the present study, probably the most important one is the position of the piston ring pack. In the optical engine, the rings are positioned about 17mm below the upper surface of the piston, whereas this distance is typically between 8 and 5mm in a production series engine. As soot potentially located in the piston top land area and the transport of soot in the blow-by gases are the topic of interest of this study, it is important to keep this difference in mind since it can potentially have an effect on the observed phenomena and especially on the interactions between the flame or the burnt gases and the ring pack. Also, the optical engine is not lubricated which means that there is no lubricant oil on the wall of the combustion chamber. This can modify the interaction between the soot and the wall in terms of the propensity of soot present in the bulk gases coming into contact with the liner and being subsequently deposited on the combustion chamber walls.
The engine was equipped with an intake heater allowing pre-heating of the fresh intake air, and with a simulated EGR (nitrogen) circuit allowing to reproduce the oxygen concentration of the intake air of real engines for varying levels of charge dilution.
An optical ring (upper part of the liner) enabled combustion visualization in the upper part of the combustion chamber as presented in Figure 3 (left). Two fused silica windows inserted into a metal ring provided the necessary optical access coupled with excellent mechanical resistance to enable part load engine operation up to 12 bar IMEP and for in-cylinder pressures up to 130 bar. The engine operating point investigated in the present study is detailed in Table 1.
For this specific investigation, the extended piston assembly was modified in such a way as to ensure an airtight seal to isolate the volume comprised between the upper section (optical piston assembly)and the lower piston (lubricated). Hereafter this volume will be referred to as the dry crankcase volume. This modification was implemented in order to enable measurements of soot concentration in the blow-by gases as well perform measurements of the optical engine blow-by mass flow rate. This cover is a metal cylinder shown in Figure 3 (right). Four ports allowed measurements on the gases located in this volume.
Fuel Injection System
The engine was equipped with a common rail fuel injection system (Bosch CP2) capable of supplying fuel at nominal rail pressures up to 1600 bar. A six-hole Bosch injector nozzle with a 145[degrees] included angle was used, mounted on a Bosch CRI2 solenoid injector body. The characteristics of the injector are listed in Table 2. The fuel used in the present study was standard European Diesel fuel whose principal physical properties are presented in Table 3.
Engine Operating Condition
One of the challenges of this study was to reach sufficiently high engine load conditions whilst being able to perform in-situ, optical measurements. It was necessary that the operating point targeted for this study be representative of a low engine speed-part load operating condition (high torque condition), where the phenomena of transport of soot-in-oil are believed to be the most problematic. This is the case of mid-high load engine loads, where high EGR rates are employed. However, because of the absence of lubricating oil in optical engines (graphite rings are used rather than a standard piston ring pack and lubricating oil, see Figure 2), optical engines are limited in terms of maximum load and engine speed. For the IFPEN optical engine, the maximum attainable engine speed is 2000rpm, and the maximum load is approximately 12-14bars. These limitations are dependent on the degree of optical access required. A summary of the characteristics of the selected operating point for this study is presented in Table 4. The corresponding in-cylinder pressure trace and Apparent Heat Release Rate (AHRR) curves are shown in Figure 4.
This section describes all the measurement procedures applied in this study, and the way these procedures have been validated.
Blow-By Mass Flow Rate Measurements
One of the four ports of the airtight housing (see Figure 3) on the extended piston assembly was used to measure engine blow-by mass flow rate. This port was connected to a calibrated sonic nozzle. The upstream and downstream pressures on both sides of the nozzle were measured. Measurement of the pressure drop and a prior calibration of the nozzle allowed determination of the mass flow rate.
The mass flow rate measurements were first calibrated in a steady-state flow condition, with the engine stationary. The crankshaft was positioned such that the intake valves were open and the exhaust valves were closed. A constant mass flow rate was imposed through the intake valves, using a sonic flow nozzle and ensuring precise control of the upstream pressure of this sonic nozzle. The piston was firstly mounted without any piston rings, so that the intake flow went through the piston ring crevices into the dry crankcase. The imposed mass flow rate was 1.15kg/h and the measured flowrate was 1.09kg/h. This yields a 5% measurement error for the steady-state flow. Subsequently, measurements were then performed with the engine in operation and the piston equipped with the full set of piston rings. The blow-by mass flow rate was measured once per engine cycle for the reference operating point described in Table 4. The blow-by mass flow rate measurements oscillated between a minimum of 0.6 and a maximum of 2kg/h, with an average value of 1.1kg/h, which represents 3.6% of the trapped air. Even if this mass flow rate is high, it is of the same order of magnitude as the blow-by mass flow rate observed (about 0.7kg/h) for similar operating points on(standard (i.e. non optical) engines. Lastly, in order to further validate the relevance of the blow-by measurements on our optical engine, a correlation between the maximum in-cylinder pressure and the blow-by mass flow rate was analyzed and is shown in Figure 5. The results show that these two values were, as expected, well correlated. This means that the observed decrease in maximum in-cylinder pressure can be attributed to a decrease in the trapped mass in the cylinder.
Based on these 3 different sets of validations, it was concluded that the selected methodology for the measurement of the blow-by mass flow rate was relevant.
Soot Concentration and Size Distribution
A Cambustion DMS500 device was used to measure the particle concentration and particle size distribution in the blow-by gases. This device was connected to another port of the airtight housing of the dry crankcase, and was also connected to the exhaust line (at 2m from the engine exhaust valves). Therefore, soot measurements were performed both in the exhaust gases and in the blow-by gases. In addition to the Cambustion DMS500 measurements, standard AVL smoke-meter measurements were performed both in the exhaust and the blow-by gases. During the validation phase, soot measurements were performed with both measurement systems, in the exhaust and in the blow-by gases.
A first validation consisted in comparing the size distribution of the soot particles measured by the DMS500 in motored (i.e. without fuel injection) and in fired (injection and combustion) conditions. The results are shown in Figure 6. This shows that in the fired operating case, the number of measured particles was about 2 orders of magnitude higher than in the motored case. This proves that the DMS500 measurements in the blow-by gases effectively measures soot, and not graphite particles which could be generated as a result of wear of the graphite piston ring pack.
Subsequently a comparison of the soot levels measured by the AVL smokemeter and the Cambustion DMS500 was performed, for both measurement locations corresponding to gases in the exhaust line and the engine blow-by. The results are presented in Figure 7. It shows that the number of particles measured by the DMS500 was proportional to the smokemeter measurements, both in the exhaust and blow-by gases. However, it must be noted that the proportionality coefficients are different: 2-[10.sup.8] for measurements performed in the exhaust line and 3 * [10.sup.7] for measurements in the blow-by gases. One explanation for this could be linked to the difference in size distribution of the particles present in the exhaust and blow-by gases. As shown in Figure 8, nucleation (small particles) and accumulation modes (large particles) are observed for the measurements performed in the exhaust line, whereas only an accumulation mode is apparent in the blow-by gases. Thus, the particle size distribution for measurements performed in the blow-by gases is shifted towards larger particles. Since the smokemeter measurements are sensitive to the projected area of the particles, for the same total number of particles, large particles will give a higher smoke value than small particles, and subsequently, a lower ratio (Value DMS500)/(Value Smokemeter).
However, the good correlation observed between AVL Smokemeter and DMS500 measurements confirms the relevance of using the DMS 500 to measure soot in the blow-by gases, despite mass-flow rates that are much lower than those observed in the exhaust line. The results of the soot measurements in the blow-by gases with the DMS500 are detailed in the Results section.
Laser-Based Measurements of Soot Deposit on the Cylinder Liner
A Laser Extinction Method (LEM, see for example [6, 7]) was used to measure soot deposits on the cylinder liner (Figure 9). The difference lies in the fact that for the present study, the LEM technique is used to measure the optical thickness of a soot deposit instead of the opacity of a soot cloud in gaseous phase. This method required optical access, which was available with the optical windows housed in the metal ring shown in Figure 3 (left). A continuous wave laser beam (wavelength of 632nm) crossed the combustion chamber through two fused silica windows. In our case, the laser beam was located 9mm below the cylinder head, in the lower part of the fused silica windows (34x12mm). The intensity of the laser beam was measured after the combustion chamber. The lower the measured intensity, the higher the absorption by the presence of soot. Since a continuous wave laser was used, the absorption of the laser beam by the soot deposit could be acquired continuously during several consecutive cycles. Figure 10 shows an example of these measured intensities for 1000 consecutive cycles. Different phases can be defined, corresponding to different timings during the cycle.
Phase I (see Figure 10) corresponds to the intake and compression strokes. During this phase, only the soot deposited on the windows during previous cycles absorbs the laser beam, since no soot is present in the combustion chamber bulk gases. One can also observe that during the "deposit only" phase, the signal is approximately constant for each cycle, meaning that the soot layer deposited on the liner does not evolve while the piston is moving. During Phase II, when the piston is near Top Dead Center (TDC), the piston completely blocks the laser beam, and the measured intensity is therefore null, as it appears on Figure 10. The duration of this phase depends on the vertical position of the laser beam, as listed in Table 5. In our case, the laser beam was located at about 9mm from the cylinder, head, which meant that LEM measurements were not possible between 30CAD before TDC and 30CAD after TDC. Finally, Phase III corresponds to the combustion and expansion strokes. Soot is present in the combustion chamber bulk gases, and therefore the measurements performed account for the presence of soot in the combustion chamber bulk gases in addition to any soot deposited on the windows, both of which will contribute to laser beam absorption. This results in a measured intensity which is lower than during Phase I, where only the soot deposit absorbs the laser beam. For the present work, only the measurement during Phase I was analyzed, since it provided an assessment of the quantity of soot deposited on the cylinder liner. For each cycle, the measured intensity was averaged over the entire duration of Phase I. This means that this method gives one value per cycle. The evolution of this value for the 1000 consecutive cycles of Figure 10 is presented in Figure 11.
Figure 10clearly shows a decrease in the collected intensity during Phase I: while the 1st engine cycle (dark blue) reveals an average intensity of approximately 1500 a.u., the last cycle (dark red) is significantly lower at about 100 a.u. This can be attributed to an increase in the optical thickness of the soot deposit on the windows which was accumulated over the 1000 consecutive fired engine cycles. This result first of all confirmed that the LEM technique was sensitive enough to monitor an increase in the soot deposit on the liner, and follow the temporal evolution. However, in order to be able to repeat these measurements, the soot deposits on the windows had to be removed after each series of measurements. This was in order to guarantee that before each measurement campaign, the windows were in the same initial state as quantified by LEM reference measurements.
A specific methodology was set up to ensure the repeatability of the measurements. For each case of study, the following steps were respected:
1. absorption measurements were performed with the engine stationary: this step allowed us to verify the stability of the laser signal.
2. absorption measurements were performed with the engine in motored operation (no injection): this provided the reference level of the LEM signal.
3. absorption measurements were performed in engine fired conditions, for 1000 consecutive engine cycles.
4. absorption measurements were performed with the engine in motored conditions (no injection), to verify measurement stability of the soot layer.
5. The liner was cleaned in order to eliminate the soot deposit.
An example of the corresponding values of the normalized LEM signal (i.e. divided by its value at first cycle) is shown in Figure 12. This figure shows that measurements performed in motored conditions (green) reveal a slightly higher absorption (i.e. lower LEM signal) than the static engine conditions (yellow). The LEM signal during the static engine phase (yellow) was about 5% higher than during the motored engine phase. However, this difference is much lower than the decrease in the LEM signal that is observed during the fired engine phase (black curve), since the LEM signal decreases from 1 to about 0.05. This decrease can be attributed to the accumulation of the soot deposit. Results confirm that the absorption signal remains constant after the last fired cycle (red), The last step consists in cleaning the windows to remove the soot layer in order to perform subsequent measurements. A laser technique was used to perform the window cleaning in order to ensure this could be done quickly and while the engine was still being motored. A diffuse, high energy (450mJ) 1064nm Nd: YAG-laser beam was directed to the window surfaces and the laser energy was sufficient to sublimate the soot deposit and clean the windows. Results shown in Figure 13 reveal the efficiency of this technique. 500 fired and 'window cleaning' engine cycles were alternatively performed. The results reveal that the LEM signal returns back to its initial level after each window cleaning sequence (500 cycles). Only a slight decrease of the LEM intensity at the beginning of the fired cycle sequence is observed: at the end of the cleaning sequence, the intensity of the LEM signal is lower than the value of 2000 a.u. which was initially measured. However, the value of the LEM intensity at the end of the cleaning phase seems to reach a plateau after the 3rd one, with intensity levels of approximately 1800 a.u. The cleaning methodology therefore seems to be efficient enough to ensure the repeatability of the LEM measurements.
3D CFD RANS computations were performed in order to shed extra light on the interactions on the phenomena of soot-wall interaction. One of the primary goals was to firstly validate the numerical simulations for this investigation and to set up specific postprocessing algorithms dedicated to the study of the different phenomena related to accumulation of particulate matter in lubricant oil. It may be pointed out that only a qualitative evolution of the soot is expected. In the following paragraphs the 3D CFD tool used is described and the setup of the model is exposed. The different dedicated post-processings used are then detailed.
Description of the 3D CFD Code
The three-dimensional simulations were performed using IFPEN's 3D CFD code , which solves the unsteady equations of a chemically reactive mixture of gases, coupled with the equations for a multi-component vaporizing fuel spray. This 3D CFD parallel (MPI) code solves the Navier-Stokes equations using an ALE-extended (Arbitrary Lagrangian Eulerian, see ,  and  for more details concerning this model and its validation) finite volume method on conformal unstructured hybrid meshes. The code uses the well-known time splitting decomposition, while the temporal integration scheme is largely implicit.
For turbulent combustion, the 3-Zone Extended Coherent Flame Model (ECFM3Z) is used . The auto-ignition prediction and the influence of chemistry on reaction rates are provided by the TKI (Tabulated Kinetics of Ignition) auto-ignition model . To obtain a realistic spray modeling during the injector needle motion, the Transient Injection Conditions (TIC) model  is used. Concerning the soot mass evaluation, the PSK (Phenomenological Soot Kinetics) soot formation model ,  is used.
Fuel Injection Modeling
The spray of the injector is discretized in parcels (set of droplets with the same characteristics). The turbulence model used was a k-[epsilon] model while the Wave/FIPA model  was used to model spray break-up and evaporation. Wall wetting by the injected fuel (liquid film) was modeled using an in-house Lagrangian model based on the O'Rourke and Amsden ,  model for droplet formation, evaporation and splashing. This model was improved using the work of Xu and Han . The modeling of liquid film dynamics takes into account the spray impact, the surface motion and the gas evolution in the vicinity of the film.
Ignition and Combustion Modeling
For compression ignited internal combustion engines, the combustion modeling relies on the combination of two models. The ECFM-3Z (a complete description is available in ) model describes the turbulent mixing in each mesh cell and provides the local thermodynamic conditions and species concentrations to the TKI auto-ignition model (more extensively described in  and ) which represents the chemical kinetics related to the combustion process. After the start of combustion, the representation of the turbulent mixing by the ECFM3Z model allows to define the local conditions in order to represent the combustion process.
In order to take into account the engine blow-by, a leakage flow was set up in the cells located in contact with the liner and the piston. This leakage flow was controlled by a defined cross section which was tuned using the motored engine in cylinder pressure measurements. The use of this model was needed in such a case to represent the optical engine. Indeed, without taking into account this blow-by phenomenon the in cylinder pressure would be overestimated. Moreover, the alternative approach would have been to impose a lower compression ratio which can enable good agreement in terms of the maximum in-cylinder pressure but this would have resulted in an overestimation of the cylinder pressure during expansion due to the corresponding decreased expansion ratio (same value as compression ratio). Figure 14 shows the evolution of the mass in the combustion chamber from the end of compression to Exhaust Valve Opening (EVO). The loss of mass computed is equivalent to that measured experimentally. The increase of mass in the fired case is due to fuel injection.
Computational Domain and Validation of Numerical Model
Wedge calculations representing only a part of the combustion chamber (considering only one spray of the injector) and modeling only the part of the engine cycle from IVC to EVO were performed with the following assumptions:
* Combustion chamber was assumed to be axisymmetric;
* The injector was located on the chamber vertical axis;
* The different phenomena were assumed to be periodic in the azimuthal direction;
* The valves and the valve recesses were neglected and the corresponding dead volumes were distributed in the combustion chamber (squish area);
* The aerodynamic motion was assumed to be a pure swirl motion of a given intensity at IVC (initial instant of the computation).
The first computations aimed to represent the reference operating condition (Table 4) using the experimental combustion chamber geometry. As a first step in terms of validation it can be observed in Figure 15 and Figure 16 that a good agreement was obtained between computational and experimental results respectively in terms of in-cylinder pressure and heat release rate. This good correlation confirmed the relevance and fidelity of the numerical tool for combustion simulations. In particular the computational results show that the different phenomena related to combustion (and subsequently soot production) are well modeled since they compare well with the experimental measurements.
Dedicated Numerical Post-Processing
The advantage of performing the 3D simulations was the ability to have access to a large number of physical parameters and to assess their spatial-temporal evolution within any zone within the combustion chamber. This was not possible with the experimental measurements. Consequently, more detailed, complementary information can be provided by the 3D simulation data regarding the different mechanisms linked to formation of soot deposits and their evolution during the engine cycle. Nevertheless, several dedicated post-processing algorithms of the numerical results were set up in order to provide valuable information on the different phenomena and mechanisms involved in the accumulation of particulate matter in lubricant oil. These specific post-processings aimed to provide comparable or additional diagnostics to those provided by the optical engine experiments.
Subdivision of the Combustion Chamber
The division of the combustion chamber into concentric rings (Figure 17) allowed monitoring and quantification of the evolution of in-cylinder soot mass (concentration and spatial location). Moreover, the extreme outer ring, that closest to the cylinder liner enabled predictions of the soot mass interacting with the wall which would likely result in deposits on the liner (see Figure 22).
Linear Exploration of Soot Field
Inspired by the LEM diagnostics set up in the experiments (see section "experimental setup"), this specific post-processing consisted of quantifying the soot mass in the combustion chamber along a line coplanar with the spray axis. The aim of this post-processing method was to provide complementary information in order to discriminate the different cases and to explain the observed trends. It should be noted that as was the case with the experimental LEM measurements, during the engine cycle, the piston may interrupt the signal for a certain crank angle window, which can be found in Table 5, depending on the line's position.
Aside from this interruption period, at each time step a numerical linear signal giving the soot density as a function of the distance to the chamber axis can be collected (Figure 19 A). Summarizing this signal along the radius gives a quantity which can be assimilated to a surface density of soot (kg/m2). Thus, at each time step this quantity can be interpreted as an opacity (inverse of a LEM gaseous signal, Figure 19 B).
One can notice that the "opacity signal" obtained depends on the vertical location of the exploration line. Figure 23 shows the evolution of "numerical opacity" as a function of the vertical location of the line (1mm=red/circle, 5mm=blue/triangle, 10mm=green/square) for different operating condition strategies.
Moreover, it is important to note that no particulate deposition model was set up in the present computations. As a consequence, only soot/wall interactions can be numerically studied without allowing to quantify the soot mass deposited.
From the reference operating conditions detailed in Table 4 several parameters were varied in order to induce changes in terms of soot generation and soot evolution in the combustion chamber. The aim of those variations is to change the soot mass located in the critical areas exposed in the paragraph "Introduction". Those variations were performed in order to check the sensitivity of the experimental and numerical tools and their ability to discriminate the different cases. The different operating conditions variations are given below, and can be split into two groups:
* variations affecting the injection process
** No pilot injection
** Start of Injection (SOI) plus or minus 5CAD
** Decrease of injection duration (1200[mu]s instead of 1600[mu]s)
* variations affecting the trapped gases.
** Increase of the simulated EGR rate (30% instead of 15%)
** Increase of the intake pressure (1.9bar instead of 1. 6bar)
For all these operating points, the relative evolutions of the LEM signals can be compared. The results are shown in Figure 20. Four repetitions of the reference point were carried out (black). This allows to estimate the standard deviation of our measurements, and then to assess whether the observed differences between operating points are significant or not. From Figure 20 the effects of the various parameters can be observed:
* Increasing the simulated EGR rate or delaying the injection will result in an increase of the soot deposit on the upper part of the cylinder liner;
* Decreasing injection duration or increasing intake pressure, advancing injection, removing the pilot injection or increasing the engine speed will result in a decrease of soot deposit.
It can be concluded that the LEM methodology is repeatable and allows to discriminate the evolution of soot deposits due to different variations in terms of operating conditions. Furthermore, the fact that the soot deposits are strongly dependent on the operating strategy indicates that slight evolutions in operating condition tuning may allow to control the soot deposit on the cylinder liner and consequently the particulate matter accumulation in lubricating oil.
These tendencies are similar to what is measured with the DMS500. Figure 19 shows the same ranking of operating points in terms of soot quantity as that observed in Figure 20, except for the increase of intake pressure (yellow line). This first tends to prove that the measurement methodologies are valid, since two different methods lead to the same ranking. It also shows that when a high amount of soot is deposited on the upper part of the combustion chamber, a high amount of soot is also measured in the blow-by gases. However, this does not give indications on which mechanism (direct transport or deposit) is predominant.
Some of the parametric variations performed experimentally around the reference operating point were carried out using the numerical approach. Figure 22 shows the evolution of the mass of soot interacting with the cylinder liner which is likely to induce deposits. Considering the global evolution leads to the following ranking: 30% EGR>Ref>No pilot>1200[mu]s. The same conclusion can be drawn from the computation of the opacity (defined in Figure 19) for the different operating points, as shown in Figure 23.
Although different in magnitude, the tendencies are the same as those observed with the LEM or DMS500 measurements. This good agreement tends to show that the qualitative soot estimation of the CFD tool gives relevant tendencies, and can therefore be used to investigate the processes of soot transport. From Figure 20, Figure 21, Figure 22 and Figure 23, one can hypothesize that a high amount of soot in the chamber leads to a high amount of soot deposited on the liner, and then to a high amount of soot found in the blow-by gases. But the mechanisms leading to the deposit and to the transport of soot in the blow-by gases still need to be identified and studied.
Mechanisms of Soot Wall Interactions
The previous section has highlighted the good agreement between CFD tools and experiments in terms of soot presence and evolution in the chamber. The idea is then to use these two complementary tools to go further into detail in the understanding of the mechanisms behind soot transport in the lubricant oil. Direct flame visualizations were experimentally performed using the window in the exhaust valve. A high-speed Photron SA5 equipped with a Tamron 70mm f/3.5 lens is viewing the combustion chamber through the window place in one of the exhaust ports, at 36 kHz (exposure time 27[mu]s). The resolution of the images is 448x320 pixels. The computed three dimensional evolution of the soot quantity can be observed from the same point of view in order to mimic this optical access. Aiming for a qualitative comparison with the experimental direct visualizations in terms of soot evolution in the combustion chamber, a numerical iso-surface of soot mass fraction colored by the temperature is tracked (as in experiments, hot areas = high soot radiation: white; cool areas = low radiation: dark). The comparison is shown in Figure 24, for different crank angle degrees (CAD) after Top Dead Center (TDC). The way to visualize the soot evolution chosen in computation is an iso-surface of soot mass fraction (at a given value). This three dimensional surface can be assimilated to an envelope containing gases with higher soot mass fraction. The holes, singularities or contusions observed are areas or volumes containing gases with a lower soot mass fraction. Nevertheless, both experimental and simulation results show similar evolutions of the soot clouds. A first soot cloud arrives near the wall about 13CAD after TDC. Then this first soot cloud undergoes a recession and tends to disappear. This recession, which will be discussed later, can be due both to cool down of the gases or to soot oxidation. After that recession, a second soot cloud appears about 45CAD after TDC. This behavior can also be observed in Figure 22.
The first increase of soot mass in interaction with the wall occurs during the beginning of combustion followed by a decrease. Then, at the end of the expansion stroke a second increase can be observed. It seems that two successive soot clouds come to interact with the wall, and therefore could be responsible for the soot deposit on the liner observed using LEM. It is quite straightforward to imagine that the first cloud corresponds to the initial fuel spray. But it is unclear why a second cloud appears. In order to investigate this phenomenon, the CFD results are further analyzed.
From this additional analysis, a scenario can be formulated. Indeed, putting together the experimental direct observations and the study of the velocity and soot numerical fields (Figure 25 gives the evolution of velocity vectors colored by the soot mass fraction) the following sequences can be observed:
* The early produced soot cloud is distributed between the squish area and the bowl (Figure 25 A). Indeed, the soot formed above the spray goes directly in the squish area toward the liner. The soot located below the spray remains in the bowl and goes toward the chamber axis. Consequently, the first breath of soot interacting with the wall is related to the spray momentum which brings the soot toward the cylinder liner (Figure 25 A & B), inducing a soot/wall interaction mainly located in the upper part of the cylinder liner, close to the head.
* The first breath of soot interacting very early with the liner is then oxidized thanks to the high temperature and the air available in the squish area (Figure 25 C & D).
* During the expansion stroke (Figure 25 C & D), the soot located in the bowl is extracted toward the cylinder liner by the inverse squish effect and reaches the liner at the end of the expansion stroke (Figure 25 E). This late interaction can be named as the second breath of soot.
These observations allow to describe the soot/liner interaction mechanism as a two phase or two breaths phenomenon: the first breath due to the spray momentum and the second due to the inverse squish effect.
One should note that the first breath is most likely to induce direct transportation of soot toward the sump through the ring pack (blow-by phenomenon) as it occurs close to the TDC when the blow-by flow rate is high.
Finally, the way the soot/liner interactions take place seems to indicate that the combustion chamber geometry (bowl shape, spray angle...) or operating strategies (SOI, existence of pilot injection, EGR ratio...) can impact the soot/wall interaction and the presence of soot in the two critical regions exposed in the "introduction" section (Top land and upper liner area). Consequently it seems that several levers may exist in order to control the accumulation of particulate matter in lubricant oil.
The present work was dedicated to the development and the validation of diagnostics for the purpose of studying and understanding the origin of particulate matter accumulation in lubricant oil. These diagnostics are based on the use of two complementary tools: optical engine test bench experiments and 3D CFD computations. The first step consisted in providing both tools with the needed means to study and quantify the different phenomena related to particulate matter transportation to the oil. The aim was to be able to observe the soot evolution in the combustion chamber, the presence of soot in two critical regions, the possible interactions between soot and cylinder liner and to measure soot mass in what would be representative of a conventional engine sump.
Consequently, on the one hand, specific experimental diagnostics were developed to measure soot in the blow-by gases and soot deposit on the liner and on the other hand, specific post-processings of 3D computation results were set up to mimic experimental diagnostics and provide additional information.
The different experimental developments and methodologies were validated, showing a good measurement repeatability and a strong sensitivity to operating condition variations. Moreover a good agreement was obtained between numerical and experimental results.
Finally, putting together the experimental and the numerical results and observations allowed to describe the soot/liner interaction mechanism as a two phase or two breaths phenomenon: the first breath due to the spray momentum and the second due to the inverse squish effect. Consequently the former, is directly dependent on the spray and the latter can be understood as a result of the interaction between the spray and the piston bowl.
Even if the goal of the present work was reached (setup of diagnostics) several limitations may be raised. Indeed, two strong differences between the present engine and a conventional one can be pointed out:
* The optical piston design is far from the piston shape of a conventional Diesel engine. As the view through the piston optical access (transparent bowl bottom) does not give valuable information on soot wall interaction, this specific piston can be replaced by an all metal more conventional shaped piston. Furthermore, overcoming the piston shape constraint will allow to study the combustion chamber geometry influence on the different mechanisms.
* The specific ring pack used in such an optical engine with a high top land is not representative of the realistic ring pack of a conventional engine. Some work has to be performed in order to improve the representativity of the ring pack.
Furthermore, some improvements must be made in terms of diagnostics in order to provide additional information to further understand the mechanisms at the origin of the particulate matter accumulation in lubricant oil:
* To be able to visualize and quantify the soot deposit on the whole height of the engine;
* To be able to distinguish the direct transportation from the scraping mechanism is mandatory in order to assess/measure the impact of the operating conditions on each mechanism separately.
Finally, even if in depth investigations are now awaited, the present work allows the description of the way the soot/liner interactions take place. The identified mechanisms indicate that the combustion chamber geometry (bowl shape, spray angle...) and the operating strategies (SOI, existence of pilot injection, EGR ratio...) are levers to act on the soot/wall interaction and the presence of soot in the two critical regions (Top land and upper liner area). Consequently it seems that several ways to control the accumulation of particulate matter in lubricant oil may exist and should be further investigated.
[1.] Lockwood, F., Zhang, Z., Choi, S., and Yu, W., "Effect of Soot Loading on the Thermal Characteristics of Diesel Engine Oils," SAE Technical Paper 2001-01-1714, 2001, doi:10.4271/2001-01-1714.
[2.] Dam, W., Willis, W., and Cooper, M., "The Impact of Additive Chemistry and Lubricant Rheology on Wear in Heavy Duty Diesel Engines," SAE Technical Paper 1999-01-3575, 1999, doi:10.4271/1999-01-3575.
[3.] Mc Geehan, J., Alexander, W., Ziemer, J., Roby, S. et al., "The Pivotal Role of Crankcase Oil in Preventing Soot Wear and Extending Filter Life in Low Emission Diesel Engines," SAE Technical Paper 1999-01-1525, 1999, doi:10.4271/1999-01-1525.
[4.] Suhre, B. and Foster, D., "In-Cylinder Soot Deposition Rates Due to Thermophoresis in a Direct Injection Diesel Engine," SAE Technical Paper 921629, 1992, doi:10.4271/921629.
[5.] Eastwood, P. (2008) Particulate Emissions from Vehicles : Wiley-PEPublishing Series.
[6.] Musculus, Mark P B; Pickett, Lyle M. (2005) Diagnostic considerations for optical laser-extinction measurements of soot in high-pressure transient combustion environments. In : Combustion and Flame, vol. 141, p. 371-391.
[7.] de Francqueville, L., Bruneaux, G., and Thirouard, B., "Soot Volume Fraction Measurements in a Gasoline Direct Injection Engine by Combined Laser Induced Incandescence and Laser Extinction Method," SAE Int. J. Engines 3(1):163-182, 2010, doi:10.4271/2010-01-0346.
[8.] Luckert P., Arnt S., Duvinage F., Kemmner M., Binz R., Stortz O., Reusch M., Braun T., Ellwanger S. The New Mercedes-Benz 4-Cylinder Diesel Engine OM654 - The Innovative Base Engine of the New Diesel Generation. 24th Aachen Colloquium Automobile and Engine Technology, pp 867-877 (2015).
[9.] Tokura, N., Terasaka, K., and Yasuhara, S., "Process through which Soot Intermixes into Lubricating Oil of a Diesel Engine with Exhaust Gas Recirculation," SAE Technical Paper 820082, 1982, doi:10.4271/820082.
[10.] Kashdan, J. and Thirouard, B., "A Comparison of Combustion and Emissions Behaviour in Optical and Metal Single-Cylinder Diesel Engines," SAE Int. J. Engines 2(1):1857-1872, 2009, doi:10.4271/2009-01-1963.
[11.] Zolver M., Klahr D., Bohbot J., Laget O., Torres A., Reactive CFD in engines with a new unstructured parallel solver, Oil & Gas Sci. and Tech. 58(1), 2003, pp 33-46.
[12.] Amsden, A.A., O'Rourke, P.J., Butler, T.D., KIVA-II, a Computer Program for Chemically Reactive Flows with Sprays, Report LA-11560-MS, Los Alamos National Laboratory,1989
[13.] Hirt, C.W., Amsden, A.A., Cook, J.L., J. Comput. Phys. pp 14, pp 227, 1974
[14.] Pracht, W.E., J. Comput. Phys. pp 17, pp 132, 1975
[15.] Colin O., Benkenida A., The 3-zones extended coherent flame model (ECFM3Z) for computing premixed/diffusion combustion, Oil & Gas Sci. and Tech., 59( 6), pp 593-609, 2004
[16.] Colin O., Pires de Cruz A., Jay S., Detailed chemistry based auto-ignition model including low temperature phenomena applied to 3-D engine calculations, Proceedings of the Combustion Institute, 30, pp 2649-2656, 2005
[17.] Beard P., Towards a predictive modelling of transient injection conditions of Diesel sprays in DID engines, Proceedings of the ILASS Americas 18th Annual conference, Irvine, USA, 2005
[18.] Martinot, S., Beard, P., Roesler, J., and Garo, A., "Comparison and Coupling of Homogeneous Reactor and Flamelet Library Soot Modeling Approaches for Diesel Combustion," SAE Technical Paper 2001-01-3684, 2001, doi:10.4271/2001-01-3684.
[19.] Jay, S., Beard, P., and Pires da Cruz, A., "Modeling Coupled Processes of CO and Soot Formation and Oxidation for Conventional and HCCI Diesel Combustion," SAE Technical Paper 2007-01-0162, 2007, doi:10.4271/2007-01-0162.
[20.] Habchi, C., Verhoeven, D., Huynh Huu, C., Lambert, L. et al., "Modeling Atomization and Break Up in High-Pressure Diesel Sprays," SAE Technical Paper 970881, 1997, doi:10.4271/970881.
[21.] O'Rourke, P. and Amsden, A., "A Particle Numerical Model for Wall Film Dynamics in Port-Injected Engines," SAE Technical Paper 961961, 1996, doi:10.4271/961961.
[22.] O'Rourke, P. and Amsden, A., "A Spray/Wall Interaction Submodel for the KIVA-3 Wall Film Model," SAE Technical Paper 2000-01-0271, 2000, doi:10.4271/2000-01-0271.
[23.] Han, Z. and Xu, Z., "Wall Film Dynamics Modeling for Impinging Sprays in Engines," SAE Technical Paper 2004-01-0099, 2004, doi:10.4271/2004-01-0099.
[24.] Buda F., Bounaceur R., Warth V., Glaude P.A., Fournet R., Battin-Leclerc F., Progress toward a unified detailed kinetic model for the autoignition of alkanes from C4 to C10 between 600 and 1200 K, Combustion and flame, Vol. 142, pp. 170-186, 2005.
IFP Energies nouvelles, Institut Carnot IFPEN Transports Energie, 1 et 4 avenue de Bois-Preau, 92852 Rueil-Malmaison Cedex - France
This work was funded by the GSM (Groupement Scientific Moteur: PSA Peugeot-Citroen, Renault SAS and IFP Energies nouvelles) and by ADEME (French Agency for Environment and Energy Control) as it was carried out within the [RAMSE.sup.3]S project framework (contract number 1382C0101).
CFD - Computational Fluid Dynamics
CAD - Crank angle degree
EGR - Exhaust Gas Recirculation
GSM - Groupement Scientifique Moteur (Research consortium involving PSA, RSA and IFPEN)
IFPEN - IFP Energies nouvelles
PSA - PSA Group (Peugeot, Citroen)
RANS - Reynolds Averaged Navier Stokes equations
RSA - Renault SA.
SOI - Start of Injection
TCO - Total Cost of Ownership
BDC - Bottom Dead Center
TDC - Top Dead Center
Olivier Laget, Louis-Marie Malbec, Julian Kashdan, and Nicolas Dronniou IFP Energies nouvelles, Institut Camot IFPEN TE
Romain Boissard PSA Peugeot Citroen
Patrick Gastaldi RENAULT SAS
Table 1. Characteristics of the IFPEN single-cylinder Diesel optical engine. Characteristics Value Number of cylinders 1 Cycle 4-Stroke Number of intake valves 2 Number of exhaust valves 1 Combustion chamber Flat optical bowl Cylinder head type Flat Displacement 500cc Bore 85 mm Stroke 88 mm Connecting Rod 145 mm Geometrical compression ratio 15.7:1 Swirl at Bottom Dead Center 1.1 Number of Valves 4 Exhaust Valve Open 34[degrees]BBDC@0.15 mm lift Exhaust Valve Close 6[degrees]BTDC@0.15 mm lift Inlet Valve Open 2[degrees]BTDC@0.15 mm lift Table 2. Characteristics of the Bosch CRI 2 injector Characteristics unit value # holes - 6 Included angle [omicron] 145 Mass flow rate / 30s (@ 100bar discharge pressure) [cm.sup.3] 395 Nozzle Tip Protrusion mm 3.35 Orifice diameter [micro]m 147 Table 3. Properties of the Diesel used in this study. Density @ 20[degrees]C kg/[m.sup.3] 831 Viscosity @ 40[degrees]C [mm.sup.2]/s 2.5 Cetane number (ISO 5165-98) - 52.1 %C-%H-%O %mass 86.1-13.3-0.6 Lower Heating Value (ASTM D 4868) MJ/kg 42.94 Table 4. Characteristics of the selected reference operating point. Characteristics Unit Value Speed Rpm 1200 IMEP Bar 10 Intake manifold pressure Bar 1.6 Equivalence ratio - 0.82 Injection pressure Bar 600 Pilot injection duration ([DSE.sup.1]) [mu]s 275 Pilot injection phasing ([SSE.sup.2]) CAD 348 Pilot injection mass Mg 1.8 Main injection duration ([DSE.sup.1]) [mu]s 1800 Main injection phasing ([SSE.sup.2]) CAD 358 Main injection mass Mg 39.4 Simulated EGR rate ([N.sub.2]) %mass 15 Equivalent real EGR rate %mass 18 Table 5. Signal interruption duration function of vertical location of the exploration line Vertical Location Signal interruption (mm from cylinder head) duration ([degrees]CA) 1 Continuous signal 5 Between 20[degrees]CA BTDC and 20[degrees]CA ATDC 10 Between 30[degrees]CA BTDC and 30[degrees]CA ATDC
|Printer friendly Cite/link Email Feedback|
|Author:||Laget, Olivier; Malbec, Louis-Marie; Kashdan, Julian; Dronniou, Nicolas; Boissard, Romain; Gastaldi,|
|Publication:||SAE International Journal of Engines|
|Date:||Dec 1, 2016|
|Previous Article:||A study isolating the effect of bore-to-stroke ratio on gasoline engine combustion chamber development.|
|Next Article:||Parameters influencing soot oxidation rates in an optical diesel engine.|