Numerical Study of Spark Plug Electrode Gap In�uence of in-cylinder Ethanol Flame Propagation

: Within the current restringing emissions regulations the trends for renewable fuel adoption, such as ethanol, have grown in the automotive industry. Besides the benefits when used as the single fuel, ethanol can also leverage the advantages in the context of hybrid vehicles by replacing the petroleum derived fuels in such configuration. In this scenario, the optimization of combustion events in internal combustion engines is paramount to not only promote high performance, but also support fuel economy. Factors such as the combustion chamber design, the positioning of the spark plug and the injector are crucial to support a successful flame propagation, avoiding misfires and decreasing knock propensity. In addition, wearing of those parts can jeopardize the occurrence of reliable and stable combustion events leading to poor emission performance, high fuel consumption and potential hardware damages due to occurrence of knocking events. This research aims to numerically analyze the effects of different spark plug electrode gaps in engine-like conditions by applying STAR-CD, a computational fluid dynamics commercial software, to mimic different configurations and operational conditions. The validation and tuning of the numerical models are conducted based on experimental tests performed in an optically accessible direct injection spark ignition engine, operating with two etha-nol-based fuels, E96W4 and E100. Thermodynamic data was simultaneously acquired and correlated with the digital UV-visible images in cycle-resolved basis. The numerical models adopted consist of 3-Zones Extended Coherent Flame and Imposed Stretch Spark Ignition Models, applied for the modeling of the combustion and the spark plug respectively.


Introduction
The current world scenario of emissions regulations has highlighted the need for reduced emissions and lower fuel consumption in internal combustion engine (ICE) vehicles, whether as the single source of power in the vehicle or within hybrid configurations [1].The viable alternative fuels is mostly needed in emerging markets, where not only the cost of electrified vehicles is prohibitive but also the power grid is deficient [2,3].Along those same lines of reduced emissions and environmentally friendly fuels are the biofuels options such as E96W4 and E100 that have attracted European and American markets in the past decade while being largely adopted in Brazilian fleets since 2000 [4,5].
Besides the political and social pressures to ban the usage of ICE, researchers have excelling to studying and implementing innovative solutions aiming to leverage the application of engines operating in lean combustion and contributing to a more efficient and less pollutant vehicle hybridization proposal [6][7][8].However, this configuration brings challenges from the technical standpoint especially within a mass production adoption.Due to the resultant instabilities in the combustion development of lean air-fuel ratios mixtures leads unsuccessful ignition (misfire) and P a g e 2 | 28 large cyclic variations and low performance.Possible solutions to overcome the challenge are usually driven towards ignition systems design improvement, as the flame kernel and the initial stages of combustion play a fundamental role in promoting effective and stable combustion progression [9,10].Literature has shown that a suitable spark plug design results in improved engine performance when operating under lean and stoichiometric conditions due to the lower occurrence of misfires, and cycle-by-cycle fluctuations [11,12].
The importance of studying the spark plug behavior in engines is related to unpredictable combustion events, such as misfires and knocking.The suitable control of the heat release guarantees reliable and stable combustion, which will imply high performance and low pollutant emissions.According to the engine load (partial or full), air-fuel ratio, and compression ratio a spark plug can behave very distinctively and lead to a range of different results regarding combustion performance and emissions [13,14].Amongst all the combustion phases, the influence of the spark plug is more relevant during the flame kernel formation stage, which can be analyzed as a two stages process.During the first stage, the mass and energy transfer processes are dominated by pressure waves and by the expansion of the plasma kernel, in the second stagelonger than the firstthe flame is self-sustained by the mass and energy transfer process through diffusion, and thermal conduction [15,16].
Two parameters are of paramount importance to determine the most suitable spark plug for a given operational condition, the electrode gap, and the discharge energy.The former is a key factor to support an efficient combustion process in internal combustion engines.The latter refers to the voltage required to provide a suitable arc that is directly proportional to the gap and related to the successful ignition of the mixture and sustainable combustion process.In other words, misfires can be caused by: (1) wide gaps, which require higher electrical voltage, and (2) narrow gaps, which may not provide enough energy for lean air-fuel mixtures to ignite.For example, the discharge energy influences the combustion phasingaffecting knock occurrenceand durationcontributing to higher thermal efficiency [17].
Whilst the causes of spark plug wearing are the aggressive environmental conditions within the combustion chamber.Factors such as the high temperatures and the intense oxidizing atmosphere contribute to the degradation of the ignition system components, most of the time increasing the temperature of the combustion chamber causing unstable and inefficient combustion and fuel consumption.The complexity of the spark plug study resides in the fact of multidisciplinary and multi-factoriality of the problem, where not only factors intrinsic to the operation are fundamental but also external factors, as mentioned above [18].
Besides the coil energy, the spark plug gap (distance between the electrodes plays a significant role in the quality of combustion and pollutant emissions.Ceper [19] conducted studies on a hydrogen-fueled engine operating with 3 different spark plug gaps.Results indicated that depending on the combination of a given spark gap and certain ignition spark timings, a significant reduction in NO emissions was obtained.Badawy et al. [16] studied the influence of the spark plug gap in the flame kernel growth area, and in the engine performance and emissions.According to his research, wider spark gaps led to a more stable combustion process (less cyclic variation).An overall increase in maximum in-cylinder pressure, turbulent flame speed, mass fraction burned rate, maximum in-cylinder temperature, and heat release rate is also observed.In terms of emissions, all these factors added to high compression ratios contribute to the increase of the NOx emissions, on the other hand, the hydrocarbon and particulate emissions tend to decrease under the same conditions.
In another study, the spark plug gap was studied with the intent of correlating spark plug outwear during its service life with exhaust gas composition.Results indicated that while the carbon dioxide content has shown slight P a g e 3 | 28 differences among the different gaps studied, the hydrocarbons and carbon monoxide emissions have increased in the wider spark gap cases [20].
In the current study, the authors applied computational fluid dynamics to analyze the flame propagation under three distinct electrode gaps in a single-cylinder engine operating with E100 and E96W4.It used the commercial software Star-CD v2021, which was initially validated based on the experiments conducted in the optically accessed singlecylinder engine [21].

Numerical Methodology
The modeling of turbulent combustion in internal combustion engines (ICE) has been largely studied for the past decades.One of these challenges is the adequate representation of the interaction between the chemical reactions and the turbulent eddies into a single term, in other words, the turbulent flame speed (sT).
The comprehensive understating of flame in-cylinder development under engine-like conditions is crucial for internal combustion engine development.Given the complexity of the task, it is fundamental to pair state-of-the-art experimental and numerical techniques to reach the required degree of sophistication and accuracy across multiple applications.
From a numerical perspective, the application of Computational Fluid Dynamics (CFD) is a very cost-effective way to provide insights into the in-cylinder flow, spray, and combustion.There is a variety of combustion models available in the literature, and in their majority, they are built based on model-specific calibration constants (tuning parameters).The consequences are mostly regards to the limitation of their large adoption by a public that requires accuracy in the results, given that tests are still required to tune the models with the appropriate constants.This research it was adopted Extended Coherent Flame Model for 3 zones (ECFM-3Z) and Imposed Stretch Spark Ignition Model (ISSIM), respectively for combustion and spark plug numerical modeling.ECFM-3Z consists of a combustion model that provides a qualitative and quantitative analysis of numerous combustion configurations (autoignition, premixed flames, diffusion flames, etc.) conferring to the model the ability to represent the different combustion events intrinsic to internal combustion engines operation.ISSIM is an ignition numerical model that derives from the flame surface density (FSD) transport equation; it provides a better representation of the flame kernel development as it takes into consideration the flame kernel interaction with the flow field [22].

Combustion: 3-zones Extended Coherent Flame (ECFM-3Z)
The 3 zones in the model refer to unmixed fuel, mixed gases, and unmixed air plus EGR zone.These zones are modeled as sub-grid quantities in situations where they are not significantly large and suitable to be resolved by a given mesh.
Equation (1) represents the mass fraction of the species (Y) in the mixed zone, which is resultant of turbulent and molecular mixing between the gases Where, m denotes the mixing zone, Zm the mean mixture fraction, and  the Dirac function.

P a g e 4 | 28
The unmixed fuel mass fractions (Yfum) and unmixed oxygen (Y02um) can be written as shown in Equations ( 2) and ( 3) Where Wm, Wf ,and W02 represents respectively the molecular mass of the mean gases, fuels, and oxygen, min is a tuning coefficient (default value of 1.0), and m is the mixing timescale.
In ECFM-3Z model, all stages of combustion, flame propagation, (auto) ignition, and post-flame/emissions are calculated based on the gases in the mixed zone.All the species in the mean space are transformed into the corresponding species in the mixed space.The mass fraction of the generic species i is then given by: Where T represents the trace species, brackets"()" refers to either (T) or non-tracer species and UM corresponds to the mass of the unmixed zone per unit of gas volume, which can be written as shown in Equation ( 5): Where Yum,I is the mass of species i in the unmixed zone per unit mass of gas, and Cx is a coefficient ratio of the unmixed to mean masses.
Equations ( 7) is used to calculate the flame surface density, defined by the flow field , and applied to calculate the flame front propagation.Equation ( 8) corresponds to the transport equation applied in STAR CD.
P a g e 5 | 28 Two terms in Equation ( 8) are crucial to capture the different flame propagation between anhydrous and hydrous ethanol,  and .The first influences the production term for flame surface density based on turbulent stretching of the flame, and the second is associated with the destruction rate for flame surface density.
After a comprehensive sensitivity analysis where the experimental thermodynamic results, such as, heat release date, in-cylinder combustion pressure, and mass fraction burned were matched with the numerical results obtained, the  and  values adopted for E96W4 were respectively 1.5 and 1.05, and for E100 were1.2 and 1.1.These coefficients were carried along all the test matrix that this research employed.
Further details and results about the methodology applied will be discussed in the following paragraphs.

Ignition Model: Imposed Stretch Spark Ignition Model (ISSIM)
Whether the critical energy required is available then an initial burnt gas mass is produced.This is represented in the Equation below: Where dle corresponds to the spark gap, x-xspk to the distance from the spark plug location xspk and the constant c0 satisfies this condition: The numerical model (3D CFD) requires the adoption of a specific value for the reaction rate progress, given by the variable   ̅̅̅̅ (, ).
Where SL is the laminar flame speed,  the flame surface density and ρ b , ρ u the density of the burnt and unburnt gas, respectively.
ISSIM aims to model the reaction rate of a growing flame kernel through the application of the flame surface density equation (FSD), employing a transition function from the start of te ignition up to the turbulent flame propagation [23].
P a g e 6 | 28

2.3.Turbulence Model: k− RNG (Renormalization Group)
This research adopts a particular approach of k- RNG model, where compressible and field forces are considered and represent better flows with high Reynolds numbers.They are based on the following equations, which represent the turbulent kinetic energy and turbulent dissipation rates.It is important to highlight the requirement of suitable wall functions in order to better represent velocity, temperature, and energy distributions in near-wall regions [24].
Turbulent Kinetic Energy: Where, P≡S ij ∂u i ∂x j (15) Where, k corresponds to the turbulent kinetic energy, p is the pressure,  is the density,  s the specific turbulent kinetic energy dissipation, C, C, C, C are specific for k- model, k- S is the surface vector, and Sj is the surface area.

2.4.Spray Model: Reitz-Diwakar
The numerical methodology adopted Reitz-Diwakar model for spray atomization.In this model, the spray cone angle is previously known and used as a boundary condition to the simulation setup.Based on the given angle, the particles initial velocity is determined.The model considers the influence of aerodynamic forces in the particle breakup as: P a g e 7 | 28 1. Bag break-up, where the non-uniform pressure field around the droplet leads to its expansion towards a low-pressure region until it disintegrates due to superficial tension stresses.
2. Stripping break-up, where shear forces are considered in the droplet surface.
The break up rate is calculated as follows: Where, Dd,stable corresponds to the characteristic diameter for the characteristic time scale tb Dd is the instantaneous droplet diameter In the Bag Break up model, the instability is determined by a critical Weber number given by: Where,   is the superficial stress coefficient.
The characteristic time associated with the calculation is given by: The criteria for Stripping break up regime is given by: We Where,   corresponds to the droplets Reynolds number.
The characteristic time scale associated with this regime is given below:

Experimental Apparatus and Methodology for Model Validation
The experimental setup consisted of the single-cylinder optical accessible engine and the high-speed camera.
Details regarding configuration and methodology are provided in the following sections.
P a g e 8 | 28 Table 1 presents the engine's operational and design parameters adopted for the experimental portion.Stoichiometric Air Fuel Ratio (AFR) 9.0 8.5 Table 3 shows the experimental conditions for all the cases studied.All the experiments were performed considering a fixed crankshaft rotational speed and load.For the first, the engine speed is controlled by the active dynamometer with an error equivalent to +1%.In terms of load, 4.8 bar was chosen based on two main factors, first a representative and typical working point at 900 rpm, and compliant with the temperature and in-cylinder pressure limitations of the optical engine.The spark advance was set at different points according to the fuel used.

Flame Propagation Calculation
This session will discuss the experimental and numerical techniques adopted for the calculation of the flame propagation speed.Ultimately, it represents the expansion rate of the flame front following the combustion reaction.It can also be explained as being a combination of fluid flow, heat conduction, chemical reaction, and mass diffusion [25].
The behavior of the flame front propagation is intrinsically dependent of the regime in which the process is occurring.The initial stages of the combustion in premixed mixtures (common in SI engines) are characterized by a laminar regime which followed by a turbulent one.The latter is directly related to the charge motion (tumble and swirl) available within the in-cylinder mixture, and it depends on combustion chamber design as well as on operational conditions, such as speed, load, and spark advance [26,27,28].
This study applied experimental acquisitions (digital images and thermodynamic data) to validate and tune the numerical model.The digital images were acquired with a Video Scope VS4-1845HS high-speed camera, and a 50mm f#4.5 Nikon lens was adopted.All tests adopted a at 5400 fps and 167.935 ns exposure timing, and two different test setups were investigated, first with the engine speed at 900 rpm and second at 1800 rpm.For the lower speed cases the acquisition resulted in 1 image/° CA, while for the higher speed the rate was 1 image/° 2CA.At least 25 cycles were obtained for each condition, where the first image corresponds to 5 crank angles before spark timing and the last to the total flame extinction.Additional details about image post-processing were discussed in a previous publication and will not be reiterated here for brevity [29].
For each test point, the acquisition of images followed the same procedure for both fuels, the engine was warmed-up for 1 minute with constant monitoring of the air-fuel ratio provided by a wide band exhaust gas oxygen sensor (accuracy +1%).Once the relative air-fuel ratio indicated the desired value (lean or stoichiometric) for the experiment the image acquisition was started and proceeded until the maximum exhaust temperature limit was reached, usually 50 to 80 cycles.The images acquisition window was limited based at the maximum in cylinder temperature and pressure that the optical engine can support.It is important to emphasize that there are multiple design differences between single-cylinder optical research and standard production engines, from the surface treatment and materials used to the higher crevices observed in the optical version, alltogether they contribute to a limited load capacity [30].
Also, the test cell conditions were controlled to provide constant inlet temperature of 21°C and ambient pressure of 0.95 bar.
To obtain the rate of flame growth for all tested fuels, enflamed areas were obtained on a cycle-by-cycle basis via thresholding and binarization.For the final post-processing, only 20 cycles (located at 50% of the acquisition window) were considered for the average flame radius post-processing.

P a g e 11 | 28
For the image post processing it was adopted ImageJ software [31,32], with Maximum entropy selected as the automatic threshold technique.In this modeling, the image is scanned until the edge of the object is determined, followed by the calculation of the enclosed area, that will support the flame radius calculation that will be discussed later [33].A sensitivity study indicated 11.2% as the threshold limit to be adopted in order to obtain the best results for enclosed flame area.Images with 8-bit grayscale images (256 intensity graduations) were used as input, where a pixel with an intensity of 0 is black, a pixel with a value of 255 is white, and everything in between is a shade of grey [32,33].The flames obtained after the numerical studies were evaluated with regard to two parameters, the average propagation speed (Si), and the Heywood Circularity Factor (HCF).
The Si is calculated based on the flame radius by using the intermediate zone as highlighted in Figure 3.The intermediate zone is used, this way the influences of the ignition moment and the cylinder wall are disregarded.The comparative analysis of flames of different fuels, and operational parameters is important to evaluate how those factors are influencing cyclic variation and combustion instabilities [27].

Numerical Setup
Figure 4 shows the mesh applied in this study, tha includes intake and exhaust ports and intake and exhaust valves.Hexahedral cells were adopted for mesh generation and the area around the spark plug was refined in 50% in order to support the spark plug studies performed in this research.The mesh size at BDC consists of 1,200,650 cells, while at TDC this size drops to 560,080 cells.

Results
The stoichiometric and lean combustion process of E100 and E96W4 were characterized regarding thermodynamic parameters and flame propagation images in a centrally mounted spray-guided DI engine by using experimental and numerical methodologies.
The results are divided into 2 sessions.The first session focus on experimental results and their application to the numerical model validation.The second session discusses the numerical results obtained for a given set of spark plug configurations evaluated in this analysis.

Validation: Experimental versus CFD
The operational parameters and the experimental outputs for in-cylinder and temperature traces were used as a baseline to the creation of a GT Power v2021.3 [35] model representing the research engine adopted.The model was applied to generate the boundary conditions for all the cases studied in the 3D Star-CD v2021 software .The crankangle resolved time histories of temperatures, pressures, densities, and mass flow rates at each point in the air-induction system, intake ports, and intake manifold as a function of engine speed and load are used as inputs to the CFD model.
Table 4 lists the cases studied for the validation of the numerical model.By using distinct engine speeds, fuels, and air-fuel ratios the accuracy of the computational fluid dynamics code and tuning parameters were confirmed.3, and similar results are obtained to the other cases.The results show a well-fitted correlation until 10aTDC when the curves start to be apart.The final phases of combustion are highly affected by the crevice volume when the flame approaches the cylinder walls, also the crevice volume, which is relatively higher in optical research engines when compared to production variants, contributes to the behavior observed.
Important to say that the crevice volume is not considered in the numerical modelling.However, the difference presented in the expanding combustion phase, after TDC, does not cause major interference in the present study, since the interest is to verify how the ignition conditions affect the initial phase of flame propagation.
P a g e 14 | 28 to point out that the simulation results are qualitatively and quantitatively close to the experimental results; the difference for a slightly higher value for the simulation results is due to the fact that the experimental result is an average of some cycles of the optical engine, and this does not reach a permanent thermal regime due to reaching an exhaust temperature limit for not to damage the quartz and sapphire parts of the engine .
It is observed that while the flame normalized area shows well correlated results between the numerical model and the experimental data, there is a remarkable difference in the flame border wrinkle when comparing both.The engine cyclic variationhighly present in low speed and low load conditionsis probably the reason, if it is present in the experimental data, but it was not modelled in the simulation.
P a g e 15 | 28 The comparison of the results was made using two distinct and independent experimental means, a qualitative and quantitative optical analysis of the flame propagation, and thermodynamic data obtained from the pressure measurement in the cylinder.Thus, there is certainty that the simulation, in general, is responding to the combustion dynamics of the optical engine and its results are representative of the intended simulations.

Thermodynamic Analysis
Table 5 present the thermodynamic results obtained by the simulation l for the mass fraction burned and the maximum in-cylinder pressure for all the cases studied.
In general, it can be pointed out that the ignition energy changes little the maximum pressure in the cylinder, when we compare the same operating point, without even showing a clear pattern of change.On the other hand, there is a considerable increase in pressure with an increase in the electrode gap.Clearly, there is an advance in combustion, as made explicit by the advance of the anchor points, for both air-fuel ratios and for all fuels.In a first analysis, it can be said that the range of energy distribution in the ignition is more relevant than the energy released, at least for the conditions investigated here.In general, the dilution effect tends to be more pronounced in E96W4 cases adopting 0.5 mm electrode spark gap, as indicated in Figure 13(b) where higher variability for HCF is shown.Besides the fact that the numerical cases did not capture cyclic variability, it is important to highlight that this would be a contributor factor to the instabilities at these early stages of the combustion, also they tend to be more pronounced in lower-speed cases.These findings were observed by other researchers when working with different fuels [38].The spark discharge energy is a contributor to overcoming the instabilities in flame propagation, as can be observed in Figures 14(a) and 14(b) where higher stability is achieved when higher energy is adopted.
P a g e 21 | 28 Figure 17 presents the average flame speed calculation obtained for the simulated cases.All the cases with lean conditions presented lower values for the flame speed when compared to the stoichiometric conditions, the trend is observed in other studies where ethanol blends were applied 10 .Also, the E100 cases indicate higher values for the average flame speed when compared to E96W4 cases.Regarding the influence of the electrode gap the results indicate that the average flame speed increases with the gap distance.A possible explanation is the higher heat losses in the cases with the lower electrode gap leading to a cooler flame and slower flame speed.The discharge energy has more pronounced influence in cases with the higher spark gap, affecting considerably the results obtained for the final velocity 36 .In summary it is inferred by these results that the influence of the spark plug design parameters and the discharge energy are the main drivers for the initial flame kernel development and influence the flame propagation [39].

Discussion and Conclusions
This study aims to evaluate numerally the in-cylinder combustion at low load and low speed considering E96W4 and E100 and different spark plug configurations.The numerical model was tuned based on the experimental study conducted initially in a spray guided DISI research engine with optical access, which provided the simultaneous acquisition of the thermodynamic data and the digital image under different operating conditions.
To provide a comprehensive study of the effectiveness of numerical models applied to spark plug and combustion modeling in internal combustion engines operating with ethanol, this research considered three distinct electrode gaps and two distinct discharge energy values.
P a g e 24 | 28 The experimental results were successfully correlated to the CFD results.Those results by themselves provide a comprehensive understanding of the differences between E100 and E96W4 flame propagation in engine-like conditions.
For both fuels, it was observed that as the spark gap increases the in-cylinder pressure increases and tends to achieve higher values at all the MFB anchor points.The trend is observed for both discharge energies analyzed.The CFD images obtained haven't shown remarkable differences in shape or flame border.
For a given spark timing, the flame radius and area showed higher growth as the spark gap increased at a similar rate for E96W4 and E100.In a real situation, where the increase of the electrode gap would be resultant of wearing, this indicates that to account for the wearing the spark timing should be adjusted to maintain the engine combustion performance.Higher in-cylinder pressures are observed for the largest gap as a result of the differences in combustion behavior caused by that.Ultimately this could lead to a reduction in engine life, and damage to other components.
The influence of the water content in E96W4 was indicated as the main reason for a similar initial flame growth in the cases where different spark plug electrode gaps were adopted.Also, the Heywood Circularity Factor obtained for the same cases indicated a more pronounced flame instability result.
Regarding the average flame speed, it was observed that the electrode gap and the discharge energy of the spark plug have significant influence in the initial flame kernel propagation for both fuels, impacting the flame propagation and possibly the knock propensity in cases with higher flame speeds due to the rapidly increase in the end gas pressure and temperature.
This research leveraged the understanding and application of E96W4 and E100 in numerical models providing correlation with experimental data.It was shown that appropriate tuning commercial numerical code can provide accurate qualitative and quantitative results to ethanol-based fuels flame propagation, as well as their thermodynamic characteristics in low load and low-speed conditions.
Future works related to flame quenching, cyclic variation, and emissions can benefit from the findings of this research work.Based on flame growth and thermodynamic parameters insights regarding combustion degradation can shed light in emissions and cold start conditions, both challenging aspects for ethanol operation given its flash point.

Figure 2
Figure 2 shows an example of one single image acquisition at the left, and the resulting images after the Maximum Entropy and Threshold methodologies are applied.The graph at right shows a final post-processed set of cases with their respective normalized flame area plotted on a CAD basis.

Figure 2 .
Figure 2. Example of flame acquisition and post-processing treatment adopted for a single image, 10 CAD AIT.

Figure 3 .
Figure 3. Normalized flame area and intermediate zone considered in the average flame propagation speed calculations.

Figure 4 .
Figure 4. Grid adopted for the numerical modeling

Table 4 .Figure 6
Figure 6(a) represents the motoring and combustion results obtained experimentally and numerically for the in-cylinder pressure for one of the cases listed in Table3, and similar results are obtained to the other cases.The results

Figure 6 (Figure 6 .
Figure 6(b) shows the correlation for the mass fraction burned duration.It can be inferred that the burning anchor points are well represented in the numerical model, supporting the extended approach to the electrode gap study in this research's scope.

Figures 10 (Figure 10 .Figure 11 .Figure 12 .Figure 13 .
Figures 10(a) and 10(b) present the post-processed results for E96W4 obtained for the Heywood CircularityFactor respectively for stoichiometric and lean-burn cases.Both cases show fluctuating results for the 0.5mm electrode gap case, while the other cases -0.7mm and 1.0mmindicate more stable values for HCF.These results can be correlated with the images shown in Figure11and 12 for the CFD results for the respective cases.Images indicate that for the flame kernel growth is slower in the cases with smaller electrode gap up to 15 CAD ASOI.However, from 20 CAD ASOI, there is no longer a remarkable difference shown among the cases.

Figure 17 .
Figure 17.Flame propagation speed (Si) versus electrode gap, obtained for the numerical cases for E100 and E96W4.

Table 2 .
Table2shows the properties for the fuels used in this study.Fuel Properties.
. Single-cylinder research engine specification and basic operational points adopted.P a g e 9 | 28